ai-orchestration-learning
Use at the end of a task in Claude Code when the work produced verified, reusable project knowledge worth saving with the MCP memory.learn tool.
Skill by ufukyuksel-dev
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/ufukyuksel-dev-ai-orchestration-learning
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: productivity, agent-building
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/ai-orchestration-learning
What's inside
- Skill name
ai-orchestration-learning- Description
- Use at the end of a task in Claude Code when the work produced verified, reusable project knowledge worth saving with the MCP memory.learn tool.
- License
- MIT
- Instructions
- 38 lines in SKILL.md · view on GitHub
Files (1)
- SKILL.md
SKILL.md (first lines)
# Project learning — Claude Code
Call `memory.learn(rootPath, candidates)` **at most once**, at task end, only when the task established knowledge that a future session would otherwise have to rediscover. No durable finding means no call. If `ai-orchestration` is unavailable or a call failed with a connection error, do not retry; just finish the task.
## Worth saving
Verified, project-specific, reusable, non-obvious:
- a decision and why (`decision`), a team convention (`convention`), a multi-step procedure (`procedure`)
- behaviour that is surprising from the code alone (`behavior`), a known failure cause (`debug_finding`)
- where a change must be made (`change_point`), how to test something (`test_address`), a data/field mapping (`field_mapping`)
## Never save
Progress notes, changed-file lists, test-pass reports, facts obvious from one read of the code, guesses, secrets, rules (rules use `rules.draft`), or anything `memory.search` already returned.
## Candidate shape
`{kind, summary (≤160 chars), content (≤700 chars, one fact), locators, appliesWhen?, limitations?, reusableFor?}`, 1–3 per call, at most 8 distinct locators in total.
- Code knowledge (`behavior`, `navigation`, `field_mapping`, `debug_finding`, `change_point`, `test_address`) needs at least one symbol or file locator copied exactly from the source you read:
- class: `{"kind":"symbol","ref":"com.example.PaymentService","path":"src/main/java/com/example/PaymentService.java"}`
- method: `{"kind":"symbol","ref":"com.example.PaymentService#approve","signature":"approve(java.lang.String)","path":"src/main/java/com/example/PaymentService.java","role":"primary_change_point"}`
- file: `{"kind":"file","ref":"src/main/resources/application.yml","role":"configuration"}`
- A directory locator is navigation only; it cannot be the sole locator.
- `decision`, `convention` and `procedure` may use `"locators": []`.
- Never invent scanner or database IDs.
## Examples
Good:
1. `{"kind":"debug_finding","summary":"Refund approval fails when the ledger lock is taken twice","content":"PaymentService.approve acquires the ledger lock and RefundJob.run acquires it again; running both in one transaction deadlocks. Refunds must go through RefundJob only.","locators":[{"kind":"symbol","ref":"com.example.PaymentService#approve","path":"src/main/java/com/example/PaymentService.java"}]}`
2. `{"kind":"decision","summary":"Invoices keep two decimal places because of the tax export","content":"Invoice amounts are rounded HALF_EVEN to two decimals because the national tax export rejects more precision (finance decision FIN-12).","locators":[]}`
3. `{"kind":"procedure","summary":"Run integration tests offline","content":"Use ./mvnw -q -o verify -Pit; the online build cannot reach the artifact proxy from CI runners.","locators":[]}`
Bad (do not send):
1. "Fixed the failing test by changing >= to ==" — a progress note.
2. "PaymentService is a Spring service" — obvious from the code.
3. "The timeout is probably configurable" — a guess, not verified.
The response reports `created`, `reused` (an existing learned record matched), `duplicates` (the gate refused an already-covered fact; nothing saved) and `rejected`. Reused and duplicate are success, not errors — do not resend. Report only what the tool returned.From ufukyuksel-dev/ai-orchestration, MIT.
More from ufukyuksel-dev/ai-orchestration
- 0.570score
ai-orchestration-memory
ufukyuksel-dev
Use when Copilot needs a prior project decision, convention, regression, or known issue through the bounded ai_orch memory search command.
SkillAFreshness: AClaude Code +9 - 0.565score
debugging-this-project
ufukyuksel-dev
Use when debugging a failure, test error, exception, or unexpected behavior in the AI Orchestration project, before reading files broadly.
SkillAFreshness: AClaude Code +9 - 0.565score
ai-orchestration-learning
ufukyuksel-dev
Use when Copilot has found durable project knowledge that should be saved with one final typed ai_orch terminal command.
SkillAFreshness: AClaude Code +9 - 0.565score
debugging-this-project
ufukyuksel-dev
Use when debugging a failure, test error, exception, or unexpected behavior in the AI Orchestration project, before reading files broadly.
SkillAFreshness: AClaude Code +9 - 0.565score
ai-orchestration-references
ufukyuksel-dev
Use when Copilot must follow a linkedReferences result or browse and read bounded shared-reference material through the read-only ai_orch reference list/read…
SkillAFreshness: AClaude Code +9 - 0.565score
ai-orchestration-rules
ufukyuksel-dev
Use when Copilot must load approved rules or the user explicitly wants a durable behavior instruction drafted, previewed, and promoted through ai_orch.
SkillAFreshness: AClaude Code +9 - 0.561score
changing-code-safely
ufukyuksel-dev
Use before editing, refactoring, renaming, or deleting code.
SkillAFreshness: AClaude Code +9 - 0.561score
ai-orchestration-learning
ufukyuksel-dev
Use at the end of a task in Codex when the work produced verified, reusable project knowledge worth saving with the MCP memory.learn tool.
SkillAFreshness: AClaude Code +9 - 0.558score
adding-an-mcp-tool
ufukyuksel-dev
Use when adding a new MCP tool or capability to the AI Orchestration server (a new tool under modules/mcp/server).
SkillAFreshness: AClaude Code +9 - 0.556score
implementation-verification
ufukyuksel-dev
Use for Copilot verification after an implementation.
SkillAFreshness: AClaude Code +9 - 0.556score
changing-code-safely
ufukyuksel-dev
Use before editing, refactoring, renaming, or deleting any class, method, endpoint, or symbol in the AI Orchestration codebase.
SkillAFreshness: AClaude Code +9 - 0.539score
ai-orchestration-jobs
ufukyuksel-dev
Use when the user explicitly asks to save or resume work ("son işi kaydet", "kaldığımız yerden devam", "save where we are", "continue the last job"), to keep…
SkillAFreshness: AClaude Code +9
Signals
- Stars
- 0
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Sep 27, 2026 (1 month ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from ufukyuksel-dev
- 0.570score
ai-orchestration-memory
ufukyuksel-dev
Use when Copilot needs a prior project decision, convention, regression, or known issue through the bounded ai_orch memory search command.
SkillAFreshness: AClaude Code +9 - 0.565score
debugging-this-project
ufukyuksel-dev
Use when debugging a failure, test error, exception, or unexpected behavior in the AI Orchestration project, before reading files broadly.
SkillAFreshness: AClaude Code +9 - 0.565score
ai-orchestration-learning
ufukyuksel-dev
Use when Copilot has found durable project knowledge that should be saved with one final typed ai_orch terminal command.
SkillAFreshness: AClaude Code +9 - 0.565score
debugging-this-project
ufukyuksel-dev
Use when debugging a failure, test error, exception, or unexpected behavior in the AI Orchestration project, before reading files broadly.
SkillAFreshness: AClaude Code +9 - 0.565score
ai-orchestration-references
ufukyuksel-dev
Use when Copilot must follow a linkedReferences result or browse and read bounded shared-reference material through the read-only ai_orch reference list/read…
SkillAFreshness: AClaude Code +9 - 0.565score
ai-orchestration-rules
ufukyuksel-dev
Use when Copilot must load approved rules or the user explicitly wants a durable behavior instruction drafted, previewed, and promoted through ai_orch.
SkillAFreshness: AClaude Code +9
Cite this page
ai-orchestration-learning skill. Agent Mart. https://agentmart.dev/skills/ufukyuksel-dev-ai-orchestration-learning (last updated 2026-10-24)