Mode d’emploi
- Copiez la commande et examinez le paquet avant de l’installer dans votre environnement agent.
- Consultez le document original pour les outils, identifiants et environnements requis.
- Décrivez les entrées et le livrable souhaité ; comparez la sortie aux sources.
npx skills add https://github.com/decisionnerd/dev-skills --skill reconAvant de commencer
- L’installation n’inclut ni abonnement au modèle, ni quota API, ni accès aux fichiers.
- Étoiles et installations reflètent l’activité, pas un test indépendant de qualité ou de sécurité.
Rendu depuis le dépôt source en conservant titres, exemples, code, tableaux, liens et images.
Recon
Situational awareness for the current repo. Recon gathers evidence, sets scope, and recommends the next skill from this pack. It does not silently implement, open PRs, or mutate GitHub unless the requested scope or a completion chain (including an issues #N handoff) calls for it.
If the user is lost (“idk”, no sense of purpose, need coaching across DocSlime / ProductFeeling / Impeccable / vendored skills), prefer idk-now instead of this skill.
Operating contract
Shared by every DecisionNerd/dev-skills skill. Canonical copy: handbook/concepts/14-operating-contract.md.
- Define done first. Before using tools, write the finish line in one or two lines: acceptance criteria (existing issue AC, BDD scenarios, tests, or contract; otherwise propose them and where they should live) and the evidence that will prove them. Re-check it before reporting done. Never report done on unverified work.
- Requested scope sets the finish line. A question ("is it ready?", "why is it broken?", "what next?") ends with the answer and a
Next:line naming the exact next invoke. An outcome request ("fix", "finish", "land", "#42") continues through the chain (diagnose → fix → test → check-readiness → merge-it) until the outcome or a real blocker. Do not end a turn with "Do you want me to…?" for in-scope, in-repo work. - Stop only for real blockers. Stop and ask only when you cannot continue without the user, or before: deleting data or unmerged work, force-push or history rewrite, changing anything outside this repository (GitHub objects, deployments, live data, production or paid resources, external services), or leaving the requested scope, unless the user's request already named that exact action. Keep the harness's permission prompts for risky commands. Otherwise keep going and put status notes in the same message as the next action.
- Ask well, once. For a genuine question use the harness's structured question tool when it has one (Claude Code:
AskUserQuestion; Codex:request_user_inputwhen the current mode supports it; T3 Code: thequestiontool) with concrete options; otherwise plain text with numbered options. Treat the answer as settled; do not re-open earlier verdicts, plans, or answers unless asked. - Fan out when work is parallel. Use subagents for independent reads (repo survey, evidence gathering, per-option research, per-area audits) and independent verification (a reviewer that did not write the change). Writes stay single-owner per path set and sequential. Brief every child with goal, done-when, constraints, must-not, and return shape; verify each result before consolidating. Use Claude Code's
Workflowtool only when orchestrating many subagents truly needs it; it is expensive. T3 Code: its delegated subagents — with optional worktree isolation for independent verification — cover this without theWorkflowtool. - Pick the model tier per child; defer to routing config. If the harness or user config already routes subagents, follow it and do not pass a model. Otherwise pick a tier per child: mechanical search or inventory → fast/cheap; implementation and evidence gathering → mid; planning, review, adversarial verification → top. Tier names differ per harness (Claude Code:
haiku/sonnet/opusorfable; T3 Code: inherit by default or pick from its live provider/model catalog). Set the choice with the harness's spawnmodelparam or agent frontmattermodel:(per-harness details:skills/agents/references/sub.md). Where the harness cannot choose, children inherit or the harness picks; say which in the status note. - Keep a checklist on long runs. For more than about five steps or work that crosses skills, keep
TASKS.mdat the repo root and tick items as they finish. Do not commit it unless the repo already tracks one. - Close every run with three headings.
Blocked on me(the one genuine question or blocker, else "none");Changed(files, commits, GitHub objects, deploys, else "nothing");Found(evidence, verdict, andNext: <exact invoke>).
Commands
| Command | Scope | Primary output |
|---|---|---|
| (none) | Infer from git logs + branch/worktree | Scope summary + ranked next-skill suggestions |
repo | Whole repository | Health/readiness recon + next skills |
issue / #N | One GitHub issue | Implementation-ready plan (see references/issue.md) |
milestone | One milestone | Progress, gaps, issue set + next skills |
help / library | — | Command + skill map |
Routing:
- No args → git-log default (below).
- First word
repo|issue|milestone|help→ that command; remainder is the target. - Bare
#123/123/ issue URL →issue. - Clear intent (“recon the launch milestone”, “scout the repo”) → map and run.
Default (no argument): git logs set scope
Run this sequence first (read-only):
git status --short --branch
git branch --show-current
git log --oneline -20
git log --oneline --decorate -10 --all
git diff --stat HEAD
git diff --stat main...HEAD 2>/dev/null || git diff --stat master...HEAD 2>/dev/nullAlso when useful: gh pr view --json number,url,title,state,baseRefName,headRefName for the current branch; parse issue numbers from branch name / recent commit messages (#123, Fixes #123).
When gh is available, fan out git, GitHub, and docs surveys to parallel read-only subagents (fast tier) and consolidate in the parent before stating the situation.
From that evidence, state:
- Scope — branch, ahead/behind, dirty files, recent themes from commits, linked issue/PR if any.
- Situation — one short paragraph (shipping? stuck? dirty WIP? main with no feature branch?).
- Next — ranked suggestions using the skill map (below), each with a one-line why and an exact invoke hint (e.g.
issues critique #42,pulls status,merge-it).
Do not ask the user to restate what git already shows unless something is ambiguous (multiple issues, wrong repo).
repo
Whole-repo recon (read-only):
- Default branch, open PR count, open issue pressure, milestone list (
ghwhen available). AGENTS.md/ CI / obvious docs gaps.- Recent
git logthemes and risk areas. - Fan out git, GitHub, and docs surveys to parallel read-only subagents (fast tier) when
ghis available; consolidate in the parent. - Suggest:
issues create,milestones plan|status,troubleshoot-app/diagnose-bugif logs imply breakage,pulls/merge-itif work is stranded on branches or release-shaped per repo policy.
Details: references/repo.md.
issue
Deep issue recon / implementation plan. Follow references/issue.md and references/planning-checklist.md.
Planning does not require native Plan Mode. Use read-only inspection in the current agent when a mode switch is unavailable, and respect any active host restrictions. A standalone planning request stops at the plan; when the request was an outcome or came from a completion workflow such as issues #N, return the plan and continue that workflow once the host permits execution. Before implementation, follow workspace ownership. If issues is not installed, use a dedicated task worktree with verified exclusive ownership, preserve the shared checkout, and carry the verified absolute path, branch, base revision, and ownership through every execution handoff and implementation result.
End with the three closing headings; Found ends with Next: <primary invoke> as a statement.
milestone
Milestone recon. Follow references/milestone.md. Summarize purpose, due date, open/closed counts, blockers; suggest milestones narrow|plan|critique|close, issues for gaps, merge-it (Stage/Ship phases) when release-shaped.
Skill map (DecisionNerd/dev-skills)
Use this map when recommending next steps. Prefer the smallest next skill that unblocks the user.
| Situation | Suggest | |||||
|---|---|---|---|---|---|---|
| Lost / idk / need vision-tied next step | idk-now (considers DocSlime, ProductFeeling, Impeccable, vendored) | |||||
| Goals/process/system/plan feel overcomplicated | `kiss audit\ | goals\ | process\ | system\ | plan\ | flow` |
| Repo split/combine/monorepo/CI/secrets (ESC) | `repos status\ | split\ | combine\ | monorepo\ | ci\ | secrets` |
| Need a new / better issue | `issues create\ | critique\ | refine\ | narrow\ | widen\ | document` |
| Organize release/version work | `milestones status\ | plan\ | critique\ | narrow` | ||
| Open or improve a PR | `pulls create\ | critique\ | status\ | refine\ | explain` | |
| Autofix → CI green → merge → close issue | merge-it | |||||
| Is the issue done enough to PR/close? | check-readiness | |||||
| Live UI / data-plane broken | troubleshoot-app (diagnoses and fixes in-repo) | |||||
| Backend / algo / API bug | diagnose-bug (diagnoses and fixes in-repo) | |||||
| Agent / LLM quality or thrash | agents slap / agents analyze (+ Langfuse via observe-it) | |||||
| Plan a repair from diagnosis | fix-it | |||||
| Need tests / coverage / BDD evidence | test-it (name regime A/B/C first) | |||||
| Need logs/metrics/traces/analytics | observe-it (regime-matched; Langfuse for C) | |||||
| Need repo docs / runbooks (surgical) | document-it — DocSlime only when altitude earned | |||||
| Need options/tradeoffs before building | research-it | |||||
| Need structure cleanup without behavior change | refactor-it | |||||
| Architecture bet needed before building | architect-it | |||||
| Agent thrashing / dumb workflow / need drain | agents slap | |||||
| Design / analyze / optimize agents or use subagents | `agents design\ | analyze\ | optimize\ | sub` | ||
| Dangling worktrees / build junk / caches / stale branches | `tidy-up scan\ | workspaces\ | artifacts\ | caches\ | deep` | |
| Staging land / production promote (repo policy) | merge-it (Stage / Ship phases) | |||||
| Unsure / no arg | this skill’s default git-log recon |
When suggesting, name the skill and a concrete command string the user (or you) can run next. One primary recommendation, then 2–3 alternates.
Output shape (default / repo / milestone)
**Scope**
- Branch: ...
- Recent work: ...
- Linked: issue/PR/milestone or none
**Situation**
<1 short paragraph>
**Recommend**
1. **Primary:** `<skill> <command> …` — <why>
2. Alternatives: ...
**Blocked on me**
<The one genuine question or blocker, else "none">
**Changed**
<Files, commits, GitHub objects touched, else "nothing">
**Found**
<Situation summary and `Next: <primary invoke>`>For issue, use the plan format in references/issue.md instead, ending with the three closing headings where Found ends with Next: <primary invoke>.
Grounding
This skill’s TTPs are grounded in current engineering baselines (DORA, GitHub Docs, Fowler/Beck, Google SRE & SWE book, OpenTelemetry, OWASP LLM / NIST AI RMF, Diátaxis — see handbook sources.md).
Situational awareness before action (SRE incident habit): read git/issue/PR state, then the smallest next move. For one issue, plan a reviewable small CL (Google eng practices).
Handbook card: handbook/practices/recon.md.

