project-knowledge
Maintain a project's canonical terms and settled decisions that future work should reuse, and record follow-ups a session discovers but does not resolve.
Skill by toy-crane
Chalk Labs
Need help deploying project-knowledge?
Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.
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/toy-crane-project-knowledge
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
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 and code. Checked Oct 26, 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/project-knowledge
What's inside
- Skill name
project-knowledge- Description
- Maintain a project's canonical terms and settled decisions that future work should reuse, and record follow-ups a session discovers but does not resolve. Use when project-specific terms are being clarified, when a choice that may constrain future work is being considered or settles, including during planning, or when a session applies a temporary workaround whose root cause stays open or observes an out-of-scope defect with evidence. Do not use for simple definitions, routine implementation details, or carrying out an already-settled decision.
- License
- MIT
- Instructions
- 114 lines in SKILL.md · view on GitHub
Files (52)
- SKILL.md
- openai.yaml · agents/
- evals.json · evals/
- metro-cache-platform-leak.md · evals/fixtures/
- CLAUDE.md · evals/fixtures/ui-contracts-claude-only/
- global.css · evals/fixtures/ui-contracts-claude-only/app/
- Button.tsx · evals/fixtures/ui-contracts-claude-only/components/ui/
- Sheet.tsx · evals/fixtures/ui-contracts-claude-only/components/ui/
- Spinner.tsx · evals/fixtures/ui-contracts-claude-only/components/ui/
- README.md · evals/fixtures/ui-contracts-claude-only/docs/decisions/
- action-progress.md · evals/fixtures/ui-contracts-claude-only/docs/decisions/
- data-ownership.md · evals/fixtures/ui-contracts-claude-only/docs/decisions/
- typography.md · evals/fixtures/ui-contracts-claude-only/docs/decisions/
- AGENTS.md · evals/fixtures/ui-contracts-design-no-route/
- DESIGN.md · evals/fixtures/ui-contracts-design-no-route/
- global.css · evals/fixtures/ui-contracts-design-no-route/app/
- Button.tsx · evals/fixtures/ui-contracts-design-no-route/components/ui/
- Sheet.tsx · evals/fixtures/ui-contracts-design-no-route/components/ui/
- Spinner.tsx · evals/fixtures/ui-contracts-design-no-route/components/ui/
- README.md · evals/fixtures/ui-contracts-design-no-route/docs/decisions/
- action-progress.md · evals/fixtures/ui-contracts-design-no-route/docs/decisions/
- data-ownership.md · evals/fixtures/ui-contracts-design-no-route/docs/decisions/
- typography.md · evals/fixtures/ui-contracts-design-no-route/docs/decisions/
- AGENTS.md · evals/fixtures/ui-contracts-no-design/
- global.css · evals/fixtures/ui-contracts-no-design/app/
- Button.tsx · evals/fixtures/ui-contracts-no-design/components/ui/
- Sheet.tsx · evals/fixtures/ui-contracts-no-design/components/ui/
- Spinner.tsx · evals/fixtures/ui-contracts-no-design/components/ui/
- README.md · evals/fixtures/ui-contracts-no-design/docs/decisions/
- action-progress.md · evals/fixtures/ui-contracts-no-design/docs/decisions/
- data-ownership.md · evals/fixtures/ui-contracts-no-design/docs/decisions/
- typography.md · evals/fixtures/ui-contracts-no-design/docs/decisions/
- AGENTS.md · evals/fixtures/ui-contracts-no-tokens/
- README.md · evals/fixtures/ui-contracts-no-tokens/docs/decisions/
- action-progress.md · evals/fixtures/ui-contracts-no-tokens/docs/decisions/
- data-ownership.md · evals/fixtures/ui-contracts-no-tokens/docs/decisions/
- typography.md · evals/fixtures/ui-contracts-no-tokens/docs/decisions/
- AGENTS.md · evals/fixtures/ui-contracts-with-design/
- DESIGN.md · evals/fixtures/ui-contracts-with-design/
- global.css · evals/fixtures/ui-contracts-with-design/app/
- Button.tsx · evals/fixtures/ui-contracts-with-design/components/ui/
- Sheet.tsx · evals/fixtures/ui-contracts-with-design/components/ui/
- Spinner.tsx · evals/fixtures/ui-contracts-with-design/components/ui/
- README.md · evals/fixtures/ui-contracts-with-design/docs/decisions/
- action-progress.md · evals/fixtures/ui-contracts-with-design/docs/decisions/
- data-ownership.md · evals/fixtures/ui-contracts-with-design/docs/decisions/
- typography.md · evals/fixtures/ui-contracts-with-design/docs/decisions/
- trigger-evals.json · evals/
- decision-contract.md · templates/
- design.md · templates/
- follow-up.md · templates/
- glossary.md · templates/
SKILL.md (first lines)
# Project Knowledge Resolve unclear project terms and preserve settled decisions so future work uses the same language and does not re-litigate the same trade-offs. ## Resolve terms Read `GLOSSARY.md` when it exists. - Resolve vague, overloaded, or conflicting project terms with the user. - Stress-test unclear relationships between domain concepts with concrete edge-case scenarios. - When a term is resolved, create `GLOSSARY.md` if needed and update it immediately using the [glossary template](./templates/glossary.md). - Keep only project terms and their current definitions in `GLOSSARY.md`. ## Preserve decisions When `docs/decisions/README.md` exists, read it and load only the subject files relevant to the current work. Record a project decision only when it is all of the following: 1. **Settled**: the choice is no longer a proposal. The user confirmed the outcome, or it was chosen under authority the user explicitly delegated for this class of decision. 2. **Reusable**: future work is likely to face the same question and should reuse this answer unless its reconsideration conditions are met. 3. **Non-obvious**: without its rationale, future work could reasonably reopen the question or choose differently. 4. **A real trade-off**: plausible alternatives existed, and preserving why they were rejected prevents the same evaluation from recurring. Require evidence that a choice was intentional before treating it as settled. Implementation or lack of objection alone is insufficient. For each qualifying decision, use the [decision contract template](./templates/decision-contract.md) to create or update its single subject file in `docs/decisions/`. Preserve only the context future work needs to apply the decision without repeating the original analysis. ## Keep the design entry current Root `DESIGN.md` is where design work starts: the project's values, the shared components in use, rejected UI patterns, and a map from each visible element to the decision contract that owns it. Contracts stay the source of every rule; `DESIGN.md` only points at them, so copy no rule text, rationale, or threshold into it. A screen contract decides how something looks or behaves on screen: color, text role, spacing, icon, component, copy, loading, disabled, or error state, or sheet, toast, or header behavior. Contracts about repository mechanics, data ownership, review process, or deployment are not screen contracts and leave `DESIGN.md` alone. A contract that sets UI review criteria may be a link target for a rejected pattern but is not a map element. When you create a screen contract, change which elements one covers, or add or remove a UI pattern that a screen or UI review-criteria contract rejects, update `DESIGN.md` in the same change using the [design entry template](./templates/design.md): - Add, correct, or remove one map line for each element the contract owns. - Add or remove one rejected-pattern line for each pattern the contract rejects that an agent would otherwise write in UI code. Process alternatives, such as enforcing a rule through lint, stay only in the contract. - Make sure both agent clients are told that work which builds or reviews screens reads `DESIGN.md` first. Put that line in `AGENTS.md`, creating the file when it is missing, and also in `CLAUDE.md` when it exists and does not already load `AGENTS.md`. Skip a file that already carries a line with that meaning. When `DESIGN.md` does not exist yet, create it at that moment: - Map every existing screen contract in the decision index, not only the current one, and list their rejected UI patterns. - Copy values and shared components once from the project's token source and shared component directory, naming each source path in its section. Omit a section whose source does not exist. After creation, leave the values and components sections to the work that changes their named sources. When an existing `DESIGN.md` uses another layout, add lines where it already keeps them instead of reformatting it. A project with no screen contract has no `DESIGN.md`. ## Record follow-ups Record a follow-up when a workaround applied in this session is temporary and its root cause stays open, or when an out-of-scope defect or suspected cause is observed with evidence. Work that closed inside the session, a guess without evidence, and a defect already fixed in the current change do not qualify. Write each qualifying item at the moment of discovery to its own `docs/follow-ups/<slug>.md` using the [follow-up template](./templates/follow-up.md), naming the slug for the symptom. An item must let a later session act without this conversation. Reporting the discovery in conversation does not preserve it. A follow-up records an open question, not a settled decision, so it does not enter `GLOSSARY.md` or a decision contract until its outcome settles on its own terms. Delete the file when the work ships or the item is promoted into a spec folder. ## Protect project truth - If code conflicts with the user's statement, `GLOSSARY.md`, or a relevant decision contract, surface the mismatch instead of choosing silently. Code shows current behavior, but not whether that behavior was intentional. - A request to align documentation with code does not by itself confirm the decision the code implies. - If relevant sources disagree about a settled decision, leave project knowledge unchanged until the intent is explicitly clarified. - Treat a term or decision as preserved only after its target file is updated.
From toy-crane/skills, MIT.
More from toy-crane/skills
- 0.609score
clean-branches
toy-crane
Clean up the local leftovers of finished work in one pass, removing worktrees and local branches whose pull request merged or closed, detached worktrees…
SkillAFreshness: A★ 3Claude Code +9 - 0.609score
resolve-follow-ups
toy-crane
Resolve evidence-backed `docs/follow-ups/*.md` through a bounded manual, Codex Scheduled task, or Claude Routine sweep.
SkillAFreshness: A★ 3Claude Code +9 - 0.609score
review-sessions
toy-crane
Review this machine's recent Claude Code and Codex session logs for work agents redo by chance every session, and report candidates to hand to a deterministic…
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
babysit-specs
toy-crane
Revise active `docs/specs/SLUG/` specs so they match what has shipped since they were written.
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
implement
toy-crane
Implement or resume settled work from a selected spec folder as one handoff bundle.
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
maintain-project-context
toy-crane
Clean, compact, and reconcile durable project context and any configured issue-tracker mirror without changing established meaning.
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
expo-dev-loop
toy-crane
Verify Expo and React Native changes in a running app with agent-device.
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
expo-smoke-test
toy-crane
Verify an Expo or React Native change together with the app's core loop on both iOS and Android using agent-device.
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
build-prototype
toy-crane
Build a complete dummy-data HTML prototype or a temporary comparison of up to three alternatives for a specific user situation.
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
merge
toy-crane
Carry the current repository change through a verified GitHub pull request merge into the requested base, or the repository's remote default branch when none…
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
pr
toy-crane
Create or reuse a ready-for-review GitHub PR for the current repository change against the requested base, or the repository's remote default branch when none…
SkillAFreshness: A★ 3Claude Code +9 - 0.564score
add-stack-context
toy-crane
Audit and establish current agent context for the technologies that define a project's stack.
SkillBFreshness: B★ 3Claude Code +9
Signals
- Stars
- 3
- Forks
- 0
- Open issues
- 0
- Last commit (this folder)
- Oct 2, 2026 (28 days ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/toy-crane/skills
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from toy-crane
- 0.609score
clean-branches
toy-crane
Clean up the local leftovers of finished work in one pass, removing worktrees and local branches whose pull request merged or closed, detached worktrees…
SkillAFreshness: A★ 3Claude Code +9 - 0.609score
review-sessions
toy-crane
Review this machine's recent Claude Code and Codex session logs for work agents redo by chance every session, and report candidates to hand to a deterministic…
SkillAFreshness: A★ 3Claude Code +9 - 0.609score
resolve-follow-ups
toy-crane
Resolve evidence-backed `docs/follow-ups/*.md` through a bounded manual, Codex Scheduled task, or Claude Routine sweep.
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
build-prototype
toy-crane
Build a complete dummy-data HTML prototype or a temporary comparison of up to three alternatives for a specific user situation.
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
implement
toy-crane
Implement or resume settled work from a selected spec folder as one handoff bundle.
SkillAFreshness: A★ 3Claude Code +9 - 0.604score
expo-smoke-test
toy-crane
Verify an Expo or React Native change together with the app's core loop on both iOS and Android using agent-device.
SkillAFreshness: A★ 3Claude Code +9
Done for you
Rather not set it up yourself?
Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.
See how we build AI agents →Read the case studies
Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.
- launches shipped
- 50+
- launches shipped
- media placements
- 500+
- media placements
- founders on every account
- 2
- founders on every account
Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block
Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.
Get project-knowledge running in your business
Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.
Cite this page
project-knowledge skill. Agent Mart. https://agentmart.dev/skills/toy-crane-project-knowledge (last updated 2026-10-24)