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 agentmods add skills/hyperlogue/r3/api-surfacenpx skills add hyperlogue/r3 --skill api-surfacegit clone --depth 1 https://github.com/hyperlogue/r3What 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 | $0.00109 | $0.04104 |
| Opus 5 | $0.00055 | $0.02052 |
| Sonnet 5 | $0.00022 | $0.00821 |
| Haiku 4.5 | $0.00011 | $0.00410 |
Grade A, and why
api-surface 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 — 244 lines — stays where its author put it; the contents beside it link to each section on GitHub.
r3's API + CLI surface
This file is the design source of truth for r3's route catalog and CLI usage —
update it here when the surface changes. The wire shapes live in
shared/types.ts (the contract itself); this is the map over them.
The HTTP/JSON API is the product, not a detail of the React client — there are
three clients (browser, CLI, agent). When you change behavior, change
shared/types.ts and keep server + CLI + web in sync.
HTTP API
Routes are served by server/index.ts behind the Host + token guards (see the
security-model skill). Highlighting runs server-side — Shiki for code,
markdown-it for .md (with per-block source-line mapping for anchoring; a
```mermaid flowchart/sequenceDiagram fence becomes a safe SVG) — shipping
tokens to the client cached by content sha, so the WASM/grammar weight never
reaches the browser.
Browse (read)
GET /api/git/status | /api/git/log | /api/git/tree | /api/diff | /api/blob —
status, paged commit history, file tree at a ref, a structured highlighted diff,
one rendered file.
Reviews
GET /api/reviews— list, queryable bysession/meta.<k>/status/repo; each row carries a livewatchingflag.POST /api/reviews— create{ kind, source, meta, title, summary }→{ id, url, review }.scratch:truecreates a scratch review and the response adds thescratchDirpath the agent drops files into;patch:'<diff>'stores a piped diff as round 1.GET /api/reviews/:id— review + feedback[] (with replies[]) + round and snapshot metas.GET …/diff[?seq=]— a diff review's rendered rounds.?seq=returns that one round ({ rounds: [that] }or{ rounds: [] }if missing); omitted returns every stored round (compat). A legacy review with no stored patches renders live as seq 0 (?seq=0or omitted). Rounds are stored wide and rendered at 3 context lines; each hunk row carriesexpandable {up,down}saying how many unchanged lines the server still HOLDS:up= the gap above that hunk,down= the gap below, set on the last hunk of each contiguous run (a body has several runs when capture itself had gaps). All-zero/absent = nothing more exists, and the client shows no expander.GET …/diff-context?file=&start=&end=&(seq= | from=&to=)[&theme=]→{ file, lines }— fill one collapsed gap.[start,end]are NEW-side line numbers, capped at 5000 rows.seqselects a diff review's stored round;from/toa files review's snapshot diff — branch on which params are present, never on a seq value, since the snapshot diff is presented as synthetic round 0 and would collide with the legacy live-render round 0. A range the source can't fully cover is a 404, never a partial fill.GET/POST/DELETE …/patches[/:seq]— list / append / drop a round.POST …/files— membership{ add?, remove? }.GET/POST/DELETE …/snapshots[/:seq],…/snapshot-diff,…/snapshot-blob— content snapshots and their derived diffs.PATCH /api/reviews/:id— edit{ status?, meta?, title?, summary?, note? }(note→meta.next_steps);DELETE /api/reviews/:id.
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 First seen · 244 lines · 109 tokens per session scan A afec03c502c5
api-surface is a skill published in the GitHub repository hyperlogue/r3 (23 stars, last pushed 7d ago), licensed MIT. It adds 109 tokens to every session and 4,104 once invoked, about $0.0005 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 2026-08-30.
Other skills, from other repositories
source-command-audit-whitepapers
Audit version freshness, FR/EN parity, and metadata quality of all whitepapers and recap cards.
self-assessment
Interactive skill assessment with personalized learning path generation.
source-command-methodology-advisor
Analyzes your codebase and asks 3 targeted questions to recommend the right AI-assisted development methodology stack.
source-command-track-mentions
Search for new online mentions of the Claude Code Ultimate Guide or Florian Bruniaux's public project portfolio and update the relevant tracker.
guide-recap
Transform CHANGELOG entries into social content (LinkedIn, Twitter/X, Newsletter, Slack) in FR + EN. Use after releases or weekly to generate ready-to-post content from guide updates.
source-command-audit-prose
Add descriptive prose to bare sections in whitepapers (FR + EN) using 9 parallel agents.