{"id":"8005917a-8649-459f-84bf-b06f4e0d1733","shortId":"U2PCHC","kind":"skill","title":"optimizing","tagline":"Use when optimizing a bundle-plugin or single skill — improving descriptions, reducing tokens, fixing audit findings, restructuring workflows, adding skills to fill gaps, or iterating on user feedback","description":"# Optimizing Bundle-Plugins\n\n## Overview\n\nOrchestrate targeted improvement of a bundle-plugin project or a single skill. Unlike a full audit, optimization focuses on goals: better triggering, lower token cost, tighter workflow chains, and feedback-driven skill refinement. This skill diagnoses issues, decides on improvements, and delegates content changes to `bundles-forge:authoring`.\n\n**Core principle:** Optimize for the agent's experience. Diagnose → decide → delegate → verify.\n\n**Skill type:** Hybrid — follow the execution flow rigidly (diagnose → decide → delegate → verify), but select targets and adapt execution strategies flexibly based on audit findings and user goals.\n\n**Announce at start:** \"I'm using the optimizing skill to improve [this project / this skill].\"\n\n## Step 1: Resolve Input & Detect Scope\n\nThe target can be a local path, a GitHub URL, or a zip file. Normalize the input to a local directory before scope detection.\n\n### Input Normalization\n\n> Edge cases & naming conventions: `bundles-forge:auditing` — `references/input-normalization.md`\n\n**This is a mandatory step — do not skip it or improvise paths.** Resolve the target to a local directory before proceeding to Scope Detection.\n\n1. **Resolve the workspace.** The workspace is `$CLAUDE_PROJECT_DIR` or `$CURSOR_PROJECT_DIR` (plugin mode), falling back to the current working directory.\n2. **Normalize the target by type:**\n   - **Local path** — use directly; no transformation needed.\n   - **GitHub URL** — parse `<owner>` and `<repo>` from the URL. Shallow-clone to `<workspace>/.bundles-forge/repos/<owner>__<repo>/` using `--depth 1 --no-checkout`, then run `git checkout`. If the directory already exists, append a `__<YYYYMMDD>` timestamp to avoid collisions. **Do not clone to `/tmp/`, `~/`, or any path outside `.bundles-forge/repos/`.**\n   - **Zip/tar.gz** — extract to `<workspace>/.bundles-forge/repos/<archive-name>/`.\n3. **Create the target subdirectory** if it does not exist.\n4. **On failure** (network error, 404, auth required, rate limit): tell the user what failed and suggest providing a local path or zip file instead. Do not silently skip or proceed with partial data.\n\nSee the canonical source for the full naming convention (version/timestamp suffixes), GitHub subdirectory URLs, and security rules.\n\n### Scope Detection\n\n**Prerequisites:** Target resolved to a local path (via Input Normalization above).\n\nAfter normalization, determine the scope from the resolved local path:\n\n| Target | How to Detect | Mode |\n|--------|--------------|------|\n| Project root | Has `skills/` directory and `package.json` | **Project optimization** — all 6 targets |\n| Single skill directory | Contains `SKILL.md` but no `skills/` subdirectory | **Skill optimization** — 3 targets + feedback iteration |\n| Single SKILL.md file | Path ends in `SKILL.md` | **Skill optimization** — 3 targets + feedback iteration |\n\n**If the target is a single skill, skip to the Skill Optimization section below.**\n\n---\n\n## Project Optimization\n\n### Process\n\n1. **Diagnose** — run audit scripts, assess skill health, detect workflow gaps\n2. **Classify & Route** — classify action type, select applicable targets\n3. **Apply** — execute selected targets, delegate content changes to authoring\n4. **Verify** — re-audit to confirm improvement\n\n### Diagnostic Tools\n\n#### Audit Script Baseline\n\nRun the quality linter to identify frontmatter issues, description anti-patterns, and broken references before manual optimization:\n\n```bash\nbundles-forge audit-skill <target-dir>        # markdown report\nbundles-forge audit-skill --json <target-dir>  # machine-readable\n```\n\nThe linter automates checks Q1-Q15 and X1-X3 from the skill quality ruleset. Focus manual effort on the subjective targets below.\n\n#### Skill Health Assessment\n\nAssess each skill across four qualitative dimensions: trigger confidence, execution clarity, end-to-end completeness, and degradation signals. See `references/optimization-decision-trees.md` for the full assessment framework and signal-to-target mapping.\n\n#### Workflow Gap Detection\n\nWhen findings reveal structural gaps (not just broken connections but missing capabilities), consider creating new skills via the CAPTURED action type. See `references/optimization-decision-trees.md` for the gap detection signals.\n\n### Routing & Classification\n\nRoute findings to targets and classify each action (FIX / DERIVED / CAPTURED) before delegating. See `references/optimization-decision-trees.md` for:\n- **Target routing table** — maps Q/W/SC findings and user signals to the 6 targets\n- **Action classification** — FIX (repair defect), DERIVED (enhance/specialize), CAPTURED (new skill for gap)\n- **Pre-delegation checklist** — classification rationale, impact analysis, scope preservation\n\n### Target 1: Skill Description Triggering\n\nThe highest-impact optimization. Descriptions are the primary mechanism for skill discovery.\n\n**Diagnosis** — identify descriptions that summarize workflow, exceed 250 characters, are too narrow/broad, or fail to start with \"Use when...\".\n\n**Decision** — draft the improved description and rationale. Use A/B eval (see below) to compare triggering accuracy before and after.\n\n**Delegation** — invoke `bundles-forge:authoring` with a precise change spec: the old description verbatim, the new description, the rationale tied to a specific diagnosis (audit finding, health assessment dimension, or user feedback), and the action classification (FIX/DERIVED). Do not ask authoring to \"improve the description\" — specify the exact change.\n\n**Guiding principle:** Use A/B eval when a change could produce regression effects — when improving one dimension might degrade another. Each eval scenario below defines its own skip conditions based on what kind of regression is possible.\n\n#### A/B Eval for Description Changes\n\nFollow `references/ab-eval-protocol.md` using the **Description Triggering** context. Compare trigger rate, false negatives, and false positives. Skip A/B if the change is purely additive.\n\n**If subagent dispatch is unavailable:** the A/B eval protocol includes a fallback section — ask the user to choose between sequential inline evaluation (read `agents/evaluator.md` and follow inline) or skipping A/B in favor of a simple verification pass.\n\n### Target 2: Content Optimization\n\n#### Token Budget\n\n> **Canonical source:** Token budgets are defined in `bundles-forge:authoring` (Token Efficiency section).\n\n**Diagnosis** — identify skills exceeding token budgets (SKILL.md body > 500 lines, bootstrap > 200 lines), duplicated content, sections that should be in `references/`.\n\n**Decision** — determine what to extract, merge, or cut. Map specific sections to their target location.\n\n**Delegation** — invoke `bundles-forge:authoring` with a section-level restructuring spec: which sections to extract (source heading → target file in references/), which content to cut (quote the specific lines), and which cross-references to add. Authoring should modify only the named sections, not rewrite the entire SKILL.md.\n\n#### Layer Assignment\n\n**Diagnosis** — verify the three-level loading structure (metadata / SKILL.md body / references) is properly layered. Identify skills where the body contains content that belongs at a different level.\n\n**Decision** — determine which sections to promote (to metadata) or demote (to references/).\n\n**Delegation** — invoke `bundles-forge:authoring` with per-section move instructions: for each section being promoted or demoted, specify the source location, the target level, and the reason (e.g. \"move lines 45-80 to references/platform-details.md because this content is only needed during platform adaptation, not on every skill load\").\n\n### Target 3: Workflow Chain Integrity\n\nConsume the `workflow-report` from `bundles-forge:auditing` (Workflow mode) to identify and fix workflow issues. If no workflow report is available, run the workflow audit first:\n\n```bash\nbundles-forge audit-workflow <target-dir>                          # full workflow audit\nbundles-forge audit-workflow --focus-skills skill-a,skill-b <root>   # focused on specific skills\n```\n\n**Fix by W-check priority:** See `references/optimization-decision-trees.md` for the full W1-W10 fix table.\n\n**After fixes — Chain A/B Eval:**\n\nFollow `references/ab-eval-protocol.md` using the **Chain Integrity** context. Use after modifying Inputs/Outputs sections, adding new skills to a chain, or when W1-W4 findings indicate integrity issues.\n\n### Target 4: Security Remediation (project only)\n\nFix security findings from `bundles-forge:auditing` Category 10.\n\n**Targets:**\n- Remove unnecessary system access from hook scripts (least privilege)\n- Scope OpenCode plugin capabilities to declared needs only\n- Remove or justify any network calls in hooks/plugins\n- Ensure agent prompts include scope constraints\n- Strip encoding tricks or obfuscated content from SKILL.md files\n\n**Process:** Run security scan first, then address findings by priority — critical before warnings, warnings before info:\n\n```bash\nbundles-forge audit-security <target-dir>\n```\n\nAlternatively, invoke `bundles-forge:auditing` for a full audit that includes security (Category 10).\n\n### Target 5: Skill & Workflow Restructuring (project only)\n\nStructural changes to achieve user goals: adding skills, replacing skills, reorganizing workflow chains, or converting skills to subagents. See `references/restructuring-operations.md` for the full step-by-step procedures (5a Adding, 5b Replacing, 5c Reorganizing, 5d Skill-to-Agent Conversion).\n\n### Target 6: Optional Component Management (project only)\n\nAdd, adjust, or migrate optional plugin components based on evolving project needs. This target handles the gap between initial scaffolding and the components a project needs as it matures.\n\n**Diagnosis** — identify signals that a component is needed. See `references/optimization-decision-trees.md` for the full signal-to-component mapping table (userConfig, MCP, LSP, output-styles, PLUGIN_DATA, path migration).\n\n**Decision** — read `bundles-forge:scaffolding` — `references/external-integration.md` for the full decision tree (CLI vs MCP, userConfig schema, PLUGIN_DATA patterns, LSP fields, output-styles format, settings.json scope).\n\n**Execution** — invoke `bundles-forge:scaffolding` using its \"Adding Optional Components\" flow. Scaffolding handles file generation, manifest updates, and inspector validation.\n\n**Verification** — after scaffolding completes, run `bundles-forge:auditing` to confirm structural integrity and security compliance (especially for new MCP servers and userConfig sensitive values).\n\n### Target 7: Deprecation and Migration (project only)\n\nCoordinate the deprecation, renaming, splitting, or merging of skills. This target ensures all references remain consistent across the project during structural changes.\n\nFor the full step-by-step process, see `references/deprecation-guide.md`.\n\n**Deprecation** — mark a skill as deprecated without removing it:\n\n1. Add `deprecated: true` and `superseded-by: <project>:<replacement>` to the skill's frontmatter\n2. Prepend the description with a deprecation notice: `\"Use when... (deprecated — use <replacement> instead)\"`\n3. Update the bootstrap routing table to note the deprecation\n4. Update cross-references in other skills' `## Integration` sections\n\n**Renaming** — change a skill's name while preserving all connections:\n\n1. Rename the directory: `skills/old-name/` → `skills/new-name/`\n2. Update frontmatter `name` field\n3. Update all cross-references (`<project>:old-name` → `<project>:new-name`) across all SKILL.md, Integration sections, and documentation\n4. Update bootstrap routing table\n6. Run `bundles-forge audit-docs` to catch any missed references\n\n**Splitting** — divide a skill into multiple focused skills:\n\n1. Design the new skill boundaries (reuse `bundles-forge:blueprinting` scenario B)\n2. Invoke `bundles-forge:scaffolding` for new skill directories\n3. Invoke `bundles-forge:authoring` to write each new skill's content\n4. Update all references to the original skill\n5. Deprecate the original (or remove if all functionality is covered)\n6. Run `bundles-forge:auditing` in workflow mode to verify chain integrity\n\n**Merging** — combine multiple skills into one:\n\n1. Design the merged skill (reuse `bundles-forge:blueprinting` scenario C)\n2. Invoke `bundles-forge:authoring` to write the merged content\n3. Deprecate the source skills\n4. Update all cross-references and routing\n5. Run `bundles-forge:auditing` in workflow mode\n\n**Platform cleanup** — after any structural change:\n\n1. Remove deprecated skill references from platform manifests (Cursor plugin.json paths)\n2. Update version-bump.json if manifest paths changed\n3. Run `bundles-forge:testing` to verify component discovery\n\n---\n\n## Single-Skill Optimization\n\nWhen the target is a single skill, run only the targets that apply at skill scope. This is auto-detected — no special flags needed.\n\n### Applicable Targets\n\n| Target | Applicable | What to Do |\n|--------|-----------|------------|\n| 1. Description Triggering | **Full** | Evaluate and improve the description's triggering accuracy |\n| 2. Content Optimization | **Full** | Check token budget, references extraction, layer assignment |\n| 3. Workflow Chain Integrity | **Partial** | Fix this skill's W9/W10 findings (Inputs/Outputs clarity, integration symmetry) |\n| 4. Security Remediation | **Partial** | Fix security issues within this skill's content |\n| 5. Skill & Workflow Restructuring | **Skip** | Project-level concern |\n| 6. Optional Component Management | **Skip** | Project-level concern |\n| 7. Deprecation and Migration | **Skip** | Project-level concern |\n| Feedback Iteration | **Full** | Process user feedback with 3-question validation |\n\n### Skill Process\n\n1. **Read target skill** — consume `skill-report` if available (or extract per-skill findings from `audit-report`)\n2. **Determine goal** — engineering optimization or feedback iteration?\n3. **Engineering path:** diagnose applicable targets (1-3, partial 4)\n4. **Feedback path:** run the Feedback Iteration process (below)\n5. **Delegate** content changes to `bundles-forge:authoring`\n6. **Verify** — run `bundles-forge:auditing` (skill mode) for post-change verification\n\n### Script Shortcuts\n\n```bash\nbundles-forge audit-skill <skill-directory>     # quality checks on single skill\n```\n\n---\n\n## Feedback Iteration\n\nProcess user feedback about a specific skill's behavior or output quality. This is a cross-cutting concern — available in both project and skill optimization modes. Use this when a user reports that a skill triggered but produced wrong results, skipped steps, or needs better wording.\n\n### Classify the Feedback\n\n| Signal | Action |\n|--------|--------|\n| \"This skill triggered but produced wrong results\" | Feedback iteration (below) |\n| \"The steps in this skill are in the wrong order\" | Feedback iteration (below) |\n| \"Description format doesn't follow conventions\" | Use optimization targets 1-2 |\n| \"Token budget exceeded across the project\" | Use optimization target 2 (project mode) |\n\n### The Feedback Process\n\n```\nReceive feedback\n  → Identify target skill\n  → If external skill: fork with `forked-` prefix before modifying\n  → Read skill, understand core goal\n  → Validate each feedback item (goal alignment, necessity, side effects)\n  → Present improvement plan to user\n  → USER CONFIRMS ← gate\n  → Copy skill to working version (<skill-name>-optimized/)\n  → Delegate changes to bundles-forge:authoring on the copy\n  → A/B eval: subagent A (original) vs subagent B (optimized) with same input\n  → Present comparison to user\n  → User adopts → replace original; User rejects → discard copy\n  → Run bundles-forge:auditing for post-change verification\n```\n\n**Validation framework** — for each feedback item, ask three questions:\n1. **Goal alignment:** Does this serve the skill's core goal, or push it toward a different purpose?\n2. **Necessity:** Without this change, does the skill have an actual defect (vs. a style preference)?\n3. **Side effects:** Could this introduce complexity creep, scope expansion, or regression?\n\n### A/B Eval for Feedback Changes\n\nFollow `references/ab-eval-protocol.md` using the **Feedback Iteration** context. Compare output quality and behavioral correctness. Skip if the feedback is about structural issues.\n\n**Rules:**\n- Never apply feedback without user confirmation of the improvement plan\n- For external skills, always fork first (prefix with `forked-`, add provenance header)\n- After all changes, invoke `bundles-forge:auditing` for post-change verification — one audit pass only (auditing reports; optimizing decides)\n\n---\n\n## Common Mistakes\n\n| Mistake | Fix |\n|---------|-----|\n| Optimizing everything at once | Pick one target, measure, improve, verify |\n| Adding MUST/ALWAYS/NEVER instead of explaining why | Explain the reasoning — agents respond to understanding |\n| Splitting skills too aggressively | Only split when there's a genuine responsibility boundary |\n| Ignoring token budget for bootstrap | Bootstrap loads every session — every word counts |\n| Applying feedback without validation | Every item goes through the 3-question framework |\n| Expanding skill scope during any optimization | Optimization should improve how well a skill fulfills its goal, not shift what the goal is. Verify after every change: does this skill still do the same thing? |\n| Running all 7 targets on a single skill | Let scope auto-detection handle it — targets 4-7 don't fully apply |\n| Rewriting entire SKILL.md instead of surgical edits | Specify section-level changes in delegation. A FIX to one heading should not trigger a full rewrite — minimize diff surface to reduce regression risk |\n| Adding third-party skills without security audit | Always run `bundles-forge:auditing` — see `references/third-party-integration.md` |\n| Adding skills without updating Integration sections | Every new connection needs symmetric `Calls` / `Called by` declarations |\n\n## Inputs\n\n- `audit-report` (optional) — findings from `bundles-forge:auditing` (full project mode). Contains per-skill breakdowns — when optimizing a single skill from a full audit, extract the relevant skill's findings from the Per-Skill Breakdown section\n- `skill-report` (optional) — findings from `bundles-forge:auditing` (skill mode). More precise input for Skill Optimization — 4-category scored report targeting a single skill\n- `workflow-report` (optional) — workflow-specific findings (W1-W11) from `bundles-forge:auditing` (workflow mode), consumed by Target 3\n- `user-feedback` (optional) — behavioral feedback about skill quality, triggering issues, or wrong output\n\n## Outputs\n\n- `optimized-skill` — improved SKILL.md content with better descriptions, reduced tokens, or fixed workflow references\n- `eval-report` (optional) — optimization record written to `.bundles-forge/evals/`, structured as:\n  - **Action type:** FIX, DERIVED, or CAPTURED\n  - **Change summary:** one sentence describing what changed and why\n  - **Diagnosis basis:** which health dimension, audit finding, or user feedback triggered this optimization\n  - **Before/after comparison:** A/B eval results, or verification pass outcome if A/B was skipped\n\n## Integration\n\n**Called by:**\n- **bundles-forge:releasing** — fix quality findings during release pipeline\n- User directly — standalone optimization of any project or skill\n\n**Calls:**\n- **bundles-forge:authoring** — all content changes (descriptions, token optimization, restructuring, third-party adaptation)\n  - Artifact: `optimized-skill` → `optimization-spec` (indirect — optimizing formulates the spec, authoring receives it as targeted change instructions)\n- **bundles-forge:scaffolding** — Platform Coverage routing for adding new platforms; Optional Component Management target for adding MCP/LSP/userConfig/output-styles\n  - Artifact: `optimized-skill` → `project-directory` (indirect — scaffolding operates on the project directory, not the optimization output)\n- **bundles-forge:auditing** — post-change verification (one pass, no loops)\n  - Artifact: `optimized-skill` → `project-directory` (indirect — auditing targets the project containing the optimized skill)\n\n**Pairs with:**\n- **bundles-forge:releasing** — after optimization, versions may need sync","tags":["optimizing","bundles","forge","odradekai","agent-skills","bundle-plugin","claude-code","claude-plugin","claude-skills","cursor-plugin"],"capabilities":["skill","source-odradekai","skill-optimizing","topic-agent-skills","topic-bundle-plugin","topic-claude-code","topic-claude-plugin","topic-claude-skills","topic-cursor-plugin"],"categories":["bundles-forge"],"synonyms":[],"warnings":[],"endpointUrl":"https://skills.sh/OdradekAI/bundles-forge/optimizing","protocol":"skill","transport":"skills-sh","auth":{"type":"none","details":{"cli":"npx skills add OdradekAI/bundles-forge","source_repo":"https://github.com/OdradekAI/bundles-forge","install_from":"skills.sh"}},"qualityScore":"0.554","qualityRationale":"deterministic score 0.55 from registry signals: · indexed on github topic:agent-skills · 208 github stars · SKILL.md body (20,339 chars)","verified":false,"liveness":"unknown","lastLivenessCheck":null,"agentReviews":{"count":0,"score_avg":null,"cost_usd_avg":null,"success_rate":null,"latency_p50_ms":null,"narrative_summary":null,"summary_updated_at":null},"enrichmentModel":"deterministic:skill-github:v1","enrichmentVersion":1,"enrichedAt":"2026-05-02T12:54:05.783Z","embedding":null,"createdAt":"2026-04-18T22:12:12.584Z","updatedAt":"2026-05-02T12:54:05.783Z","lastSeenAt":"2026-05-02T12:54:05.783Z","tsv":"'-2':2067 '-3':1931 '-7':2409 '-80':1049 '/.bundles-forge/repos':253,291 '/evals':2607 '/repos':287 '/tmp':279 '1':142,206,256,438,662,1506,1562,1618,1692,1743,1807,1896,1930,2066,2178 '10':1192,1271 '2':229,449,869,1519,1568,1631,1704,1754,1819,1916,2077,2196 '200':899 '250':686 '3':292,404,417,458,1067,1532,1573,1641,1715,1761,1830,1891,1924,2212,2355,2565 '4':302,468,1178,1542,1592,1654,1720,1845,1933,1934,2408,2536 '404':307 '45':1048 '5':1273,1662,1728,1857,1943 '500':896 '5a':1307 '5b':1309 '5c':1311 '5d':1313 '6':391,637,1320,1597,1673,1866,1952 '7':1459,1875,2394 'a/b':706,770,803,824,837,860,1148,2135,2224,2640,2648 'access':1197 'accuraci':713,1818 'achiev':1282 'across':548,1481,1585,2071 'action':453,599,617,639,752,2033,2610 'actual':2206 'ad':21,1162,1285,1308,1420,2308,2446,2462,2716,2724 'adapt':115,1060,2688 'add':961,1326,1507,2270 'addit':830 'address':1240 'adjust':1327 'adopt':2152 'agent':92,1220,1317,2317 'agents/evaluator.md':854 'aggress':2324 'align':2107,2180 'alreadi':267 'altern':1257 'alway':2264,2454 'analysi':658 'announc':126 'anoth':785 'anti':491 'anti-pattern':490 'append':269 'appli':459,1787,2252,2346,2413 'applic':456,1800,1803,1928 'artifact':2689,2726,2756 'ask':757,844,2175 'assess':443,544,545,569,745 'assign':975,1829 'audit':17,52,121,180,441,472,478,504,512,742,1080,1098,1105,1109,1114,1190,1255,1262,1266,1441,1603,1678,1733,1914,1958,1973,2163,2280,2287,2290,2453,2459,2479,2487,2504,2527,2559,2630,2747,2764 'audit-doc':1602 'audit-report':1913,2478 'audit-secur':1254 'audit-skil':503,511,1972 'audit-workflow':1104,1113 'auth':308 'author':86,467,722,758,884,929,962,1021,1646,1709,1951,2131,2677,2701 'auto':1794,2403 'auto-detect':1793,2402 'autom':520 'avail':1094,1905,2001 'avoid':273 'b':1124,1630,2142 'back':223 'base':119,795,1333 'baselin':480 'bash':499,1100,1250,1968 'basi':2626 'before/after':2638 'behavior':1990,2240,2570 'belong':999 'better':57,2027,2588 'blueprint':1628,1701 'bodi':895,986,995 'bootstrap':898,1535,1594,2338,2339 'boundari':1623,2333 'breakdown':2495,2516 'broken':494,587 'budget':873,877,893,1825,2069,2336 'bundl':7,33,42,84,178,285,501,509,720,882,927,1019,1078,1102,1111,1188,1252,1260,1387,1415,1439,1600,1626,1634,1644,1676,1699,1707,1731,1764,1949,1956,1970,2129,2161,2278,2457,2485,2525,2557,2605,2655,2675,2709,2745,2775 'bundle-plugin':6,32,41 'bundles-forg':83,177,284,500,508,719,881,926,1018,1077,1101,1110,1187,1251,1259,1386,1414,1438,1599,1625,1633,1643,1675,1698,1706,1730,1763,1948,1955,1969,2128,2160,2277,2456,2484,2524,2556,2604,2654,2674,2708,2744,2774 'c':1703 'call':1216,2473,2474,2652,2673 'canon':338,874 'capabl':591,1206 'captur':598,620,646,2615 'case':174 'catch':1606 'categori':1191,1270,2537 'chain':64,1069,1147,1154,1167,1291,1684,1832 'chang':81,465,726,766,774,807,827,1280,1486,1553,1742,1760,1946,1964,2126,2167,2200,2228,2275,2284,2383,2425,2616,2622,2680,2706,2750 'charact':687 'check':521,1133,1823,1976 'checklist':654 'checkout':259,263 'choos':848 'clariti':555,1842 'classif':609,640,655,753 'classifi':450,452,615,2029 'claud':213 'cleanup':1738 'cli':1396 'clone':251,277 'collis':274 'combin':1687 'common':2294 'compar':711,815,2236 'comparison':2148,2639 'complet':560,1436 'complex':2218 'complianc':1448 'compon':1322,1332,1348,1360,1371,1422,1769,1868,2720 'concern':1865,1874,1883,2000 'condit':794 'confid':553 'confirm':474,1443,2117,2256 'connect':588,1561,2470 'consid':592 'consist':1480 'constraint':1224 'consum':1071,1900,2562 'contain':396,996,2491,2768 'content':80,464,870,902,948,997,1054,1230,1653,1714,1820,1856,1945,2586,2679 'context':814,1156,2235 'convent':176,344,2062 'convers':1318 'convert':1293 'coordin':1465 'copi':2119,2134,2158 'core':87,2100,2187 'correct':2241 'cost':61 'could':775,2215 'count':2345 'cover':1672 'coverag':2713 'creat':293,593 'creep':2219 'critic':1244 'cross':958,1545,1577,1724,1998 'cross-cut':1997 'cross-refer':957,1544,1576,1723 'current':226 'cursor':217,1751 'cut':916,950,1999 'data':335,1381,1402 'decid':75,96,108,2293 'decis':698,909,1004,1384,1394 'declar':1208,2476 'defect':643,2207 'defin':790,879 'degrad':562,784 'deleg':79,97,109,463,622,653,717,924,1016,1944,2125,2427 'demot':1013,1034 'deprec':1460,1467,1497,1502,1508,1525,1529,1541,1663,1716,1745,1876 'depth':255 'deriv':619,644,2613 'describ':2620 'descript':13,489,664,671,681,702,730,734,762,806,812,1522,1808,1815,2057,2589,2681 'design':1619,1693 'detect':145,170,205,354,379,446,579,606,1795,2404 'determin':368,910,1005,1917 'diagnos':73,95,107,439,1927 'diagnosi':679,741,888,976,1355,2625 'diagnost':476 'diff':2440 'differ':1002,2194 'dimens':551,746,782,2629 'dir':215,219 'direct':238,2665 'directori':167,200,228,266,385,395,1565,1640,2732,2739,2762 'discard':2157 'discoveri':678,1770 'dispatch':833 'divid':1611 'doc':1604 'document':1591 'doesn':2059 'draft':699 'driven':68 'duplic':901 'e.g':1045 'edg':173 'edit':2420 'effect':778,2110,2214 'effici':886 'effort':536 'encod':1226 'end':412,557,559 'end-to-end':556 'engin':1919,1925 'enhance/specialize':645 'ensur':1219,1476 'entir':972,2415 'error':306 'especi':1449 'eval':707,771,787,804,838,1149,2136,2225,2597,2641 'eval-report':2596 'evalu':852,1811 'everi':1063,2341,2343,2350,2382,2468 'everyth':2299 'evolv':1335 'exact':765 'exceed':685,891,2070 'execut':104,116,460,554,1412 'exist':268,301 'expand':2358 'expans':2221 'experi':94 'explain':2312,2314 'extern':2089,2262 'extract':289,913,940,1827,1907,2505 'fail':316,692 'failur':304 'fall':222 'fallback':842 'fals':818,821 'favor':862 'feedback':30,67,406,419,749,1884,1889,1922,1935,1939,1980,1984,2031,2041,2054,2081,2084,2104,2173,2227,2233,2245,2253,2347,2568,2571,2634 'feedback-driven':66 'field':1405,1572 'file':160,325,410,944,1233,1426 'fill':24 'find':18,122,581,611,631,743,1173,1185,1241,1840,1911,2482,2510,2522,2551,2631,2660 'first':1099,1238,2266 'fix':16,618,641,1086,1129,1143,1146,1183,1835,1849,2297,2429,2593,2612,2658 'fix/derived':754 'flag':1798 'flexibl':118 'flow':105,1423 'focus':54,534,1117,1125,1616 'focus-skil':1116 'follow':102,808,856,1150,2061,2229 'forg':85,179,286,502,510,721,883,928,1020,1079,1103,1112,1189,1253,1261,1388,1416,1440,1601,1627,1635,1645,1677,1700,1708,1732,1765,1950,1957,1971,2130,2162,2279,2458,2486,2526,2558,2606,2656,2676,2710,2746,2776 'fork':2091,2093,2265,2269 'format':1409,2058 'formul':2698 'four':549 'framework':570,2170,2357 'frontmatt':487,1518,1570 'fulfil':2371 'full':51,342,568,1107,1139,1265,1301,1367,1393,1489,1810,1822,1886,2437,2488,2503 'fulli':2412 'function':1670 'gap':25,448,578,584,605,650,1342 'gate':2118 'generat':1427 'genuin':2331 'git':262 'github':155,242,347 'goal':56,125,1284,1918,2101,2106,2179,2188,2373,2378 'goe':2352 'guid':767 'handl':1340,1425,2405 'head':942,2432 'header':2272 'health':445,543,744,2628 'highest':668 'highest-impact':667 'hook':1199 'hooks/plugins':1218 'hybrid':101 'identifi':486,680,889,991,1084,1356,2085 'ignor':2334 'impact':657,669 'improv':12,38,77,136,475,701,760,780,1813,2112,2259,2306,2366,2584 'improvis':192 'includ':840,1222,1268 'indic':1174 'indirect':2696,2733,2763 'info':1249 'initi':1344 'inlin':851,857 'input':144,163,171,363,2146,2477,2532 'inputs/outputs':1160,1841 'inspector':1431 'instead':326,1531,2310,2417 'instruct':1027,2707 'integr':1070,1155,1175,1445,1550,1588,1685,1833,1843,2466,2651 'introduc':2217 'invok':718,925,1017,1258,1413,1632,1642,1705,2276 'issu':74,488,1088,1176,1851,2249,2576 'item':2105,2174,2351 'iter':27,407,420,1885,1923,1940,1981,2042,2055,2234 'json':514 'justifi':1213 'kind':798 'layer':974,990,1828 'least':1201 'let':2400 'level':934,981,1003,1041,1864,1873,1882,2424 'limit':311 'line':897,900,954,1047 'linter':484,519 'load':982,1065,2340 'local':152,166,199,235,321,360,374 'locat':923,1038 'loop':2755 'lower':59 'lsp':1376,1404 'm':130 'machin':516 'machine-read':515 'manag':1323,1869,2721 'mandatori':185 'manifest':1428,1750,1758 'manual':497,535 'map':576,629,917,1372 'mark':1498 'markdown':506 'matur':1354 'may':2781 'mcp':1375,1398,1452 'mcp/lsp/userconfig/output-styles':2725 'measur':2305 'mechan':675 'merg':914,1471,1686,1695,1713 'metadata':984,1011 'might':783 'migrat':1329,1383,1462,1878 'minim':2439 'miss':590,1608 'mistak':2295,2296 'mode':221,380,1082,1681,1736,1960,2008,2079,2490,2529,2561 'modifi':964,1159,2096 'move':1026,1046 'multipl':1615,1688 'must/always/never':2309 'name':175,343,967,1557,1571,1581,1584 'narrow/broad':690 'necess':2108,2197 'need':241,1057,1209,1337,1351,1362,1799,2026,2471,2782 'negat':819 'network':305,1215 'never':2251 'new':594,647,733,1163,1451,1583,1621,1638,1650,2469,2717 'new-nam':1582 'no-checkout':257 'normal':161,172,230,364,367 'note':1539 'notic':1526 'obfusc':1229 'old':729,1580 'old-nam':1579 'one':781,1691,2286,2303,2431,2618,2752 'opencod':1204 'oper':2735 'optim':1,4,31,53,89,133,389,403,416,432,436,498,670,871,1774,1821,1920,2007,2064,2075,2124,2143,2292,2298,2363,2364,2497,2535,2582,2600,2637,2667,2683,2691,2694,2697,2728,2742,2758,2770,2779 'optimization-spec':2693 'optimized-skil':2581,2690,2727,2757 'option':1321,1330,1421,1867,2481,2521,2547,2569,2599,2719 'orchestr':36 'order':2053 'origin':1660,1665,2139,2154 'outcom':2646 'output':1378,1407,1992,2237,2579,2580,2743 'output-styl':1377,1406 'outsid':283 'overview':35 'package.json':387 'pair':2772 'pars':244 'parti':2449,2687 'partial':334,1834,1848,1932 'pass':867,2288,2645,2753 'path':153,193,236,282,322,361,375,411,1382,1753,1759,1926,1936 'pattern':492,1403 'per':1024,1909,2493,2514 'per-sect':1023 'per-skil':1908,2492,2513 'pick':2302 'pipelin':2663 'plan':2113,2260 'platform':1059,1737,1749,2712,2718 'plugin':8,34,43,220,1205,1331,1380,1401 'plugin.json':1752 'posit':822 'possibl':802 'post':1963,2166,2283,2749 'post-chang':1962,2165,2282,2748 'pre':652 'pre-deleg':651 'precis':725,2531 'prefer':2211 'prefix':2094,2267 'prepend':1520 'prerequisit':355 'present':2111,2147 'preserv':660,1559 'primari':674 'principl':88,768 'prioriti':1134,1243 'privileg':1202 'procedur':1306 'proceed':202,332 'process':437,1234,1494,1887,1895,1941,1982,2082 'produc':776,2020,2038 'project':44,138,214,218,381,388,435,1181,1277,1324,1336,1350,1463,1483,1863,1872,1881,2004,2073,2078,2489,2670,2731,2738,2761,2767 'project-directori':2730,2760 'project-level':1862,1871,1880 'promot':1009,1032 'prompt':1221 'proper':989 'protocol':839 'proven':2271 'provid':319 'pure':829 'purpos':2195 'push':2190 'q/w/sc':630 'q1':523 'q1-q15':522 'q15':524 'qualit':550 'qualiti':483,532,1975,1993,2238,2574,2659 'question':1892,2177,2356 'quot':951 'rate':310,817 'rational':656,704,736 're':471 're-audit':470 'read':853,1385,1897,2097 'readabl':517 'reason':1044,2316 'receiv':2083,2702 'record':2601 'reduc':14,2443,2590 'refer':495,908,946,959,987,1015,1478,1546,1578,1609,1657,1725,1747,1826,2595 'references/ab-eval-protocol.md':809,1151,2230 'references/deprecation-guide.md':1496 'references/external-integration.md':1390 'references/input-normalization.md':181 'references/optimization-decision-trees.md':565,602,624,1136,1364 'references/platform-details.md':1051 'references/restructuring-operations.md':1298 'references/third-party-integration.md':2461 'refin':70 'regress':777,800,2223,2444 'reject':2156 'releas':2657,2662,2777 'relev':2507 'remain':1479 'remedi':1180,1847 'remov':1194,1211,1504,1667,1744 'renam':1468,1552,1563 'reorgan':1289,1312 'repair':642 'replac':1287,1310,2153 'report':507,1075,1092,1903,1915,2014,2291,2480,2520,2539,2546,2598 'requir':309 'resolv':143,194,207,357,373 'respond':2318 'respons':2332 'restructur':19,935,1276,1860,2684 'result':2022,2040,2642 'reus':1624,1697 'reveal':582 'rewrit':970,2414,2438 'rigid':106 'risk':2445 'root':382 'rout':451,608,610,627,1536,1595,1727,2714 'rule':352,2250 'ruleset':533 'run':261,440,481,1095,1235,1437,1598,1674,1729,1762,1782,1937,1954,2159,2392,2455 'scaffold':1345,1389,1417,1424,1435,1636,2711,2734 'scan':1237 'scenario':788,1629,1702 'schema':1400 'scope':146,169,204,353,370,659,1203,1223,1411,1790,2220,2360,2401 'score':2538 'script':442,479,1200,1966 'section':433,843,887,903,919,933,938,968,1007,1025,1030,1161,1551,1589,2423,2467,2517 'section-level':932,2422 'secur':351,1179,1184,1236,1256,1269,1447,1846,1850,2452 'see':336,564,601,623,708,1135,1297,1363,1495,2460 'select':112,455,461 'sensit':1456 'sentenc':2619 'sequenti':850 'serv':2183 'server':1453 'session':2342 'settings.json':1410 'shallow':250 'shallow-clon':249 'shift':2375 'shortcut':1967 'side':2109,2213 'signal':563,573,607,634,1357,1369,2032 'signal-to-compon':1368 'signal-to-target':572 'silent':329 'simpl':865 'singl':10,47,393,408,426,1772,1780,1978,2398,2499,2542 'single-skil':1771 'skill':11,22,48,69,72,99,134,140,384,394,400,402,415,427,431,444,505,513,531,542,547,595,648,663,677,890,992,1064,1118,1120,1123,1128,1164,1274,1286,1288,1294,1315,1473,1500,1516,1549,1555,1613,1617,1622,1639,1651,1661,1689,1696,1719,1746,1773,1781,1789,1837,1854,1858,1894,1899,1902,1910,1959,1974,1979,1988,2006,2017,2035,2048,2087,2090,2098,2120,2185,2203,2263,2322,2359,2370,2386,2399,2450,2463,2494,2500,2508,2515,2519,2528,2534,2543,2573,2583,2672,2692,2729,2759,2771 'skill-a':1119 'skill-b':1122 'skill-optimizing' 'skill-report':1901,2518 'skill-to-ag':1314 'skill.md':397,409,414,894,973,985,1232,1587,2416,2585 'skills/new-name':1567 'skills/old-name':1566 'skip':189,330,428,793,823,859,1861,1870,1879,2023,2242,2650 'sourc':339,875,941,1037,1718 'source-odradekai' 'spec':727,936,2695,2700 'special':1797 'specif':740,918,953,1127,1987,2550 'specifi':763,1035,2421 'split':1469,1610,2321,2326 'standalon':2666 'start':128,694 'step':141,186,1303,1305,1491,1493,2024,2045 'step-by-step':1302,1490 'still':2387 'strategi':117 'strip':1225 'structur':583,983,1279,1444,1485,1741,2248,2608 'style':1379,1408,2210 'subag':832,1296,2137,2141 'subdirectori':296,348,401 'subject':539 'suffix':346 'suggest':318 'summar':683 'summari':2617 'supersed':1512 'superseded-bi':1511 'surfac':2441 'surgic':2419 'symmetr':2472 'symmetri':1844 'sync':2783 'system':1196 'tabl':628,1144,1373,1537,1596 'target':37,113,148,196,232,295,356,376,392,405,418,423,457,462,540,575,613,626,638,661,868,922,943,1040,1066,1177,1193,1272,1319,1339,1458,1475,1777,1785,1801,1802,1898,1929,2065,2076,2086,2304,2395,2407,2540,2564,2705,2722,2765 'tell':312 'test':1766 'thing':2391 'third':2448,2686 'third-parti':2447,2685 'three':980,2176 'three-level':979 'tie':737 'tighter':62 'timestamp':271 'token':15,60,872,876,885,892,1824,2068,2335,2591,2682 'tool':477 'topic-agent-skills' 'topic-bundle-plugin' 'topic-claude-code' 'topic-claude-plugin' 'topic-claude-skills' 'topic-cursor-plugin' 'toward':2192 'transform':240 'tree':1395 'trick':1227 'trigger':58,552,665,712,813,816,1809,1817,2018,2036,2435,2575,2635 'true':1509 'type':100,234,454,600,2611 'unavail':835 'understand':2099,2320 'unlik':49 'unnecessari':1195 'updat':1429,1533,1543,1569,1574,1593,1655,1721,1755,2465 'url':156,243,248,349 'use':2,131,237,254,696,705,769,810,1152,1157,1418,1527,1530,2009,2063,2074,2231 'user':29,124,314,633,748,846,1283,1888,1983,2013,2115,2116,2150,2151,2155,2255,2567,2633,2664 'user-feedback':2566 'userconfig':1374,1399,1455 'valid':1432,1893,2102,2169,2349 'valu':1457 'verbatim':731 'verif':866,1433,1965,2168,2285,2644,2751 'verifi':98,110,469,977,1683,1768,1953,2307,2380 'version':2123,2780 'version-bump.json':1756 'version/timestamp':345 'via':362,596 'vs':1397,2140,2208 'w':1132 'w-check':1131 'w1':1141,1171,2553 'w1-w10':1140 'w1-w11':2552 'w1-w4':1170 'w10':1142 'w11':2554 'w4':1172 'w9/w10':1839 'warn':1246,1247 'well':2368 'within':1852 'without':1503,2198,2254,2348,2451,2464 'word':2028,2344 'work':227,2122 'workflow':20,63,447,577,684,1068,1074,1081,1087,1091,1097,1106,1108,1115,1275,1290,1680,1735,1831,1859,2545,2549,2560,2594 'workflow-report':1073,2544 'workflow-specif':2548 'workspac':209,211 'write':1648,1711 'written':2602 'wrong':2021,2039,2052,2578 'x1':527 'x1-x3':526 'x3':528 'zip':159,324 'zip/tar.gz':288","prices":[{"id":"5c93e64b-7a44-4fab-ac9a-7364c0dc9888","listingId":"8005917a-8649-459f-84bf-b06f4e0d1733","amountUsd":"0","unit":"free","nativeCurrency":null,"nativeAmount":null,"chain":null,"payTo":null,"paymentMethod":"skill-free","isPrimary":true,"details":{"org":"OdradekAI","category":"bundles-forge","install_from":"skills.sh"},"createdAt":"2026-04-18T22:12:12.584Z"}],"sources":[{"listingId":"8005917a-8649-459f-84bf-b06f4e0d1733","source":"github","sourceId":"OdradekAI/bundles-forge/optimizing","sourceUrl":"https://github.com/OdradekAI/bundles-forge/tree/main/skills/optimizing","isPrimary":false,"firstSeenAt":"2026-04-18T22:12:12.584Z","lastSeenAt":"2026-05-02T12:54:05.783Z"}],"details":{"listingId":"8005917a-8649-459f-84bf-b06f4e0d1733","quickStartSnippet":null,"exampleRequest":null,"exampleResponse":null,"schema":null,"openapiUrl":null,"agentsTxtUrl":null,"citations":[],"useCases":[],"bestFor":[],"notFor":[],"kindDetails":{"org":"OdradekAI","slug":"optimizing","github":{"repo":"OdradekAI/bundles-forge","stars":208,"topics":["agent-skills","bundle-plugin","claude-code","claude-plugin","claude-skills","cursor-plugin"],"license":"apache-2.0","html_url":"https://github.com/OdradekAI/bundles-forge","pushed_at":"2026-04-27T02:13:58Z","description":"An agentic skills framework & bundle-plugin engineering toolkit that works.","skill_md_sha":"16ede68b999bbaf7712dffb7a34619d1e81cd5b0","skill_md_path":"skills/optimizing/SKILL.md","default_branch":"main","skill_tree_url":"https://github.com/OdradekAI/bundles-forge/tree/main/skills/optimizing"},"layout":"multi","source":"github","category":"bundles-forge","frontmatter":{"name":"optimizing","description":"Use when optimizing a bundle-plugin or single skill — improving descriptions, reducing tokens, fixing audit findings, restructuring workflows, adding skills to fill gaps, or iterating on user feedback"},"skills_sh_url":"https://skills.sh/OdradekAI/bundles-forge/optimizing"},"updatedAt":"2026-05-02T12:54:05.783Z"}}