coding-agent-docs-optimization skill (samber): docs a coding agent can use
Makes SDK and API docs something a coding agent can integrate from on its own, and measures it with a repeatable cold run.
- Installs
- 12.4K
- on skills.sh
- Stars
- 7
- on GitHub
- Files
- 6
- besides SKILL.md
- Updated
- Nov 2026
- last repo commit
coding-agent-docs-optimization makes the documentation half of agent experience work: it prepares your SDK, API or protocol docs so a coding agent can integrate from them on the first attempt, and proves it with a repeatable run. It checks crawler access first, runs a baseline cold run per task, inventories seven machine-readable surfaces, rewrites the pages failed runs touched, annotates the spec and re-runs the same task set.
It triggers on "agent-readable docs", "llms.txt", "agents inventing API calls" or "could an agent wire up our SDK from the docs". It bundles no scripts. SKILL.md says to fetch pages with each agent user-agent string you care about when Claude can make outbound requests, and to fix the documentation, never the prompt.
Good for
- Vendors whose users report an agent inventing methods that do not exist
- Maintainers asked to add llms.txt
- API teams checking whether an agent could wire up their SDK from the docs alone
Install coding-agent-docs-optimization
Claude Code
- 1
Copy this prompt.
Install the coding-agent-docs-optimization skill in Claude Code. Run this command: /plugin marketplace add samber/developer-relations-skills /plugin install developer-relations-skills@developer-relations-skills Then check that it loaded and tell me how to use it.
- 2
Paste it into Claude Code. It shows the change and asks before running it.
The README recommends installing the whole collection because its skills reference each other, and coding-agent-docs-optimization names other skills in the collection to hand work to. The whole-collection routes add all 93 skills in the repo (counted from the git tree); the repo holds no hooks, agents, commands or MCP server files, and 14 of its skills bundle a script. Pick one route per agent.
From the project README. Source.
What's in the skill
When it loads
Makes SDK, API or protocol documentation something a coding agent can integrate from unattended - crawler access, machine-readable entry points (llms.txt, markdown endpoints, OpenAPI and JSON Schema files), self-contained copy-paste-safe pages, and a measured first-attempt agent success rate. Use whenever the user mentions agent-readable or AI-ready docs, agent experience or AX, llms.txt or llms-full.txt, docs for coding agents, agents inventing API calls that do not exist, or asks whether an agent could wire up their SDK from the docs alone - even if they never say "agent". Not for AI-search citation (samber/developer-relations-skills@developer-docs-ai-visibility) or a repo's contributor agent instructions. For human docs architecture use samber/developer-relations-skills@developer-docs-structure-audit.
Sections in SKILL.md
- Scope
- Interview
- Workflow
- Which surface to build first
- Exit criteria
- Page rules
- Spec rules
- Ownership shapes the plan
- Invocation examples
- Output shape
- Failure modes
- References
Key facts
- Skill version
- 1.0.3 (SKILL.md metadata)
- Installs on skills.sh
- 12,428 (Nov 2, 2026)
- Last repo commit
- Nov 1, 2026
- Audits on skills.sh
- Gen Agent Trust Hub: pass, Socket: pass, Snyk: pass
What it includes
Asks 10 questions; the 5-10 integration jobs that matter most define the task set.
InterviewFetches a page and the index with each agent user-agent string and records status codes.
Step 1: Access checkUses a fresh agent, published sources only and logs every corrective turn.
A fresh agent restricted to published surfaces, one run per task, logging every corrective turn.
Step 2: Baseline cold runMarks seven surfaces present, partial or missing.
Marks index file, full-corpus bundle, version labels, markdown endpoints, annotated specs, SDK doc comments and terminal paths as present, partial or missing.
Step 3: Surface inventoryApplies ten rules, such as standalone pages, pinned versions and complete snippets.
Ten rules such as standalone pages, pinned versions and complete runnable snippets.
Page rulesApplies five rules, from stable operationIds to per-language samples.
Five rules: stable operationIds, closed value spaces, curation before tool conversion, split summary and description, per-language samples.
Spec rulesRequires four criteria, including a first-attempt success target.
Four: first-attempt success against a target, zero invented API surface on top tasks, at most three documents per routine task and fewer corrective turns.
Exit criteriaDelivers six parts, from baseline scores to re-run scores.
Six parts from baseline scores to re-run scores.
Audit report
From the official docs
Permissions
Low risk
coding-agent-docs-optimization is audit instructions with no scripts, and it may fetch your docs pages and save notes to devrel-context.md.
What it can do
- Read your data
- Create and change things
- Delete things
- Read-only mode
Watch out for
It may fetch your docs with several user agents
SKILL.md's first workflow step says that, if Claude can make outbound HTTP requests, it fetches a page and the index with each agent user-agent string you care about and records the status codes. Our advice: run it against sites you own. github.com
It may save notes to devrel-context.md
SKILL.md tells Claude to keep task and project context in devrel-context.md when memory lives in a file, or in the memory system your tool already uses. Check the file before you commit it. github.com
The whole-set install adds 93 skills, 14 with scripts
The README says to install every skill. The git tree has 93 skill folders, and 14 of them hold .py or .sh files. This skill bundles no scripts, and the tree holds no hooks, agents or MCP config files. github.com
Technical details
| Turning on read-only | SKILL.md describes no read-only mode. |
|---|---|
| Talks to | Your docs site and index URLs, when Claude can make outbound HTTP requests |
| Sign-in and keys | SKILL.md names no account, key or token. |
| Scopes |
|
Questions
What does coding-agent-docs-optimization do?
It audits and improves your SDK or API documentation for coding agents. It scores how often an agent completes your key integration tasks unattended, then fixes the pages and spec fields where it failed.
How do I install it?
Use the install box at the top. The README says to install every skill in the repo, because the skills point to each other. The box shows the whole-set route for each app.
Does it just write an llms.txt?
No. SKILL.md says to inventory first, generate the index from your navigation source, advertise it in three places and say plainly that the file alone changes nothing measurable.
Does it fetch my site with different user agents?
If Claude can make outbound HTTP requests, SKILL.md tells it to fetch a page and the index with each agent user-agent string you care about and record the status codes. Otherwise it has the docs owner run those requests.
Is it safe to install?
The skill bundles no scripts. Its fetches go to your docs site, and SKILL.md notes that admitting or refusing a crawler is a policy call usually signed off outside the docs team. The whole-set install adds 93 skills, 14 of which bundle scripts.
How is it different from developer-docs-ai-visibility?
This skill is about agents integrating from your docs. SKILL.md routes getting cited by AI search to samber/developer-relations-skills@developer-docs-ai-visibility and human docs architecture to developer-docs-structure-audit.
Information
updated Nov 2, 2026
- Repository
- samber/developer-relations-skills
- Folder
skills/coding-agent-docs-optimization
Sources
Docs, issue trackers and discussions this page is built from.
Show all 3Hide
- README · github.com
- coding-agent-docs-optimization SKILL.md · github.com
- LICENSE · github.com