Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add OutlineDriven/outline-driven-development --skill docs-and-adrsgit clone --depth 1 https://github.com/OutlineDriven/outline-driven-developmentWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/outlinedriven/outline-driven-development/docs-and-adrs)<a href="https://agentmods.dev/skills/outlinedriven/outline-driven-development/docs-and-adrs"><img src="https://agentmods.dev/badge/skills/outlinedriven/outline-driven-development/docs-and-adrs/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/outlinedriven/outline-driven-development/docs-and-adrs"><img src="https://agentmods.dev/badge/skills/outlinedriven/outline-driven-development/docs-and-adrs.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00046 | $0.01649 |
| Opus 5 | $0.00023 | $0.00825 |
| Sonnet 5 | $0.00009 | $0.00330 |
| Haiku 4.5 | $0.00005 | $0.00165 |
Grade A, and why
docs-and-adrs scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 50 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Docs and ADRs
Contract
| Field | Bound contract |
|---|---|
| Trigger | Making an architectural decision, changing a public API, shipping a user-facing feature, capturing context for future engineers and agents, or a codebase term resolving |
| Authority | Reversible local: writes only the documentation artifacts listed under Side effect, inside the current project working tree; rollback is version control. No remote mutation. Nothing is committed, staged, published, or pushed. |
| Side effect | ADR, README, API/JSDoc/OpenAPI docs, inline comments, changelog and agent-rules files; deletions are limited to commented-out code |
| Done | ADR exists for each significant decision, README/API/inline gotchas accurate, no commented-out code, agent rules current, each resolved codebase term routed to domain-modeling |
Inputs
- Required: the decision, API change, feature, or context that fired the trigger, with the rationale and constraints the user can supply.
- Optional: existing ADR directory, README,
CHANGELOG.md, OpenAPI spec, and agent-rules file; where one is absent, create it at the location the Procedure names. - Optional: the repository's documentation convention (ADR directory, numbering, extension, markup, heading set), detected by inspection; an established convention overrides the Procedure's defaults.
- Dates, rejected alternatives, and constraints come only from the user or the repository; never invent them.
Procedure
- Bound scope before any write: enumerate only the decisions, APIs, and features named by the trigger. Do not document code whose meaning is obvious from reading it, write comments restating what code already says, or document throwaway prototypes. Done when: the scope list names only items from the trigger and nothing invented.
- Inspect the repository for an established documentation convention: existing ADRs, project instructions, ADR tooling config. Match the existing location, extension, markup, numbering, and heading set; when evidence conflicts, surface the conflict instead of introducing a second scheme. Apply the defaults below only when no convention exists. Done when: the convention is matched, or the default layout is selected with any conflict surfaced.
- Route resolved codebase terms to domain-modeling. When a codebase term resolves during this pass, hand the term, its project definition, and its near-synonyms to domain-modeling, which owns
CONTEXT.mdand its entry schema. Do not write toCONTEXT.mdhere. Done when: each resolved term is routed to domain-modeling, or no term resolved. - For each significant decision, write one ADR. Follow
references/adrs.mdfor the qualifying conditions that gate whether a decision warrants an ADR, thedocs/decisions/storage location with sequential numbering, and the template with its optional fields. Read the existing ADR directory to confirm the next number before writing. Done when: an ADR file exists at the location and format the reference specifies (or the detected convention's path) for each qualifying decision, with sequential numbering continuing the existing sequence. - Manage the ADR lifecycle in place. Mark an ADR recording a decision taken in this session as
accepted. When a later ADR reverses an earlier one, set the old ADR's status tosuperseded by NNNN. Never delete an ADR file. Done when: each ADR's status field reflects its lifecycle position. - Inline comments: write only why-comments that explain the constraint, trade-off, or trap the code cannot show. Replace what-comments (
i++; // increment i) with why-comments (i++; // retry budget: the upstream limiter drops the first burst per connection). Done when: every comment in the changed surface is a why-comment; no what-comments remain. - Document each known trap as a gotcha comment at the exact place a future engineer or agent would hit it. State the trigger and the reason, and cross-reference the governing ADR by number where one exists (
// NOTE: call flush() before close(); close() silently drops buffered records otherwise. See ADR 0007.). Delete commented-out code on this pass. Report a TODO comment that has sat for weeks as stale instead of leaving it as documentation. Done when: each known trap has a gotcha comment at its code site and no commented-out code remains in the changed surface. - API documentation: for every public API function added or changed, write JSDoc with its TypeScript parameter and return types, thrown errors, and a usage example. For every REST endpoint added or changed, add or update its OpenAPI/Swagger entry in the project's OpenAPI spec, including path, method, parameters, and response schema. Done when: every public API function in scope has typed JSDoc and every REST endpoint in scope has an OpenAPI entry.
- README: when the project has no README or its README is stale relative to this work, update it to cover quick start, commands, an architecture overview linking to ADRs, and contributing, preserving existing correct content. Done when: README covers quick start, commands, architecture overview, and contributing.
- Changelog: when shipping a feature that changes user-facing behavior, add a Keep-a-Changelog-style entry at the top of
CHANGELOG.md(create## [Unreleased]when absent) under one ofAdded,Changed,Deprecated,Removed,Fixed,Security, with the issue or PR reference where one exists. Done when: a changelog entry exists under the correct section for each user-facing behavior change. - Keep agent-facing documentation current in the same pass. Put agent conventions in the agent-rules file (
CLAUDE.mdorAGENTS.md). Keep spec files updated so agents build the right thing. Use ADRs to record why past decisions were made so agents do not re-decide them. Place inline gotchas where agents will encounter them. Done when: agent-rules files and spec files reflect current conventions and no settled decision lacks a discoverable rationale. - Stop rather than widen scope: never expand the pass into documenting the whole codebase, and never write an artifact whose content would have to be invented. Done when: no artifact was written whose content had to be invented and no out-of-trigger code was documented.
What ships with it
6 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2d ago Changed · +2 lines · -18 tokens per session 3c1f8f1139f4
- 5d ago First seen · 48 lines · 64 tokens per session scan A 3f85790288b6
docs-and-adrs is a skill published in the GitHub repository OutlineDriven/outline-driven-development (52 stars, last pushed 3d ago), licensed Apache-2.0. It adds 46 tokens to every session and 1,649 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-09-03.
Other skills, from other repositories
audit-project
Run an iterative multi-agent code audit until critical and high findings are resolved. Use when the user says "audit my code", "find all the bugs", "deep code audit", "iterative review", or "review until clean".
duet
Use when the user invokes /duet, says "pair with me", or faces aesthetic, architectural, or irreversible decisions.
goal-prompt-drafting
Use when asked to draft copy-ready /goal objectives for long-running agents; returns one normalized one-line objective with measurable end state, grounded proof, easy-out invariants, a stop clause, and a Missing list. Not for source or remote-system changes.
handoff-prompt
Use when the user asks for a handoff, delegation, or clipboard-ready prompt for another agent: a standalone path-free prompt copied to the clipboard, confirmed by title. Not for session-snapshot briefs — use handoff; never remote, credential, publish, deploy, or irreversible.
publish-branch
Use when asked to publish the checked-out branch: commit and push it on whatever branch it is, the default branch included. Not for creating branches, PRs, force pushes, or pushing any other branch; when the request excludes the default branch, use commit-push-current.
snippet-image-rendering
Use when the user explicitly names snipgrapher and wants code rendered to a polished PNG, SVG, or WebP at an explicit local path. Probes the installed CLI for supported flags and writes only the named image file. Not for other renderers, publishing, or remote actions.