jakevin
afa5e6046c
* refactor: consolidate 6 skills into 3, remove mechanical commands Replaces opencli-oneshot / opencli-explorer / opencli-browser / opencli-usage with a single opencli-adapter-author skill that takes the AI agent end-to-end: site recon, API discovery, field decoding, adapter coding, and `opencli browser verify`. Removes the mechanical commands (`explore`, `synthesize`, `generate`, `cascade`, `record`) and their src/tests — they were codegen scaffolding meant for agents, which the new skill handles more flexibly via `opencli browser` primitives. Skill highlights: - Top-level decision tree + 12-step runbook - 5 site patterns (SPA / SSR / JSONP / Token / Streaming) - 5-layer API discovery (network → initial state → bundle → token → interceptor) - Field decode playbook (self-explanatory → codes → sort-key comparison) - Output design guide (columns, types, order, ≤15 per adapter) - Two-layer site memory: in-repo seeds for eastmoney/xueqiu/bilibili/tonghuashun plus local `~/.opencli/sites/<site>/` runtime workspace Kept skills: opencli-autofix (now points to adapter-author for rewrites), smart-search. Kept primitives: `browser *`, `doctor`, `list`, `validate`, `verify`, `<site> <cmd>`, `plugin *`, `completion`. No backward compatibility shims. Full test suite (1605 tests) passes. * review fixes: honest coverage, hard memory-hit path, typo, stale docs - site-memory hit path no longer jumps to writing adapter; forces Step 5 endpoint re-verification + Step 7 field check, and 30-day expiry - site-memory.md now specifies exact schemas for endpoints.json / field-map.json / notes.md / fixtures + write-back timing rules - coverage-matrix.md marks unverified patterns as 🟡 with an evidence section citing coingecko dry run + PR #1091 eastmoney + bilibili - eastmoney seed typo: resolveSecids -> resolveSecid (and splitSymbols) - docs/developer/ai-workflow.md rewritten to teach the adapter-author skill + opencli browser * primitives (dropped generate/synthesize/ cascade/explore references) - ts-adapter.md, getting-started.md, CHANGELOG.md:87 updated to point at opencli-adapter-author * fix(ci): resync package-lock + drop stale built-in list reference - Regenerate package-lock.json to restore @emnapi/core + @emnapi/runtime entries that got dropped during the rebase — `npm ci` was failing on all CI jobs (build / audit / docs-build / bun-test / unit-test) - docs/guide/getting-started.md: built-in list dropped `explore`, now reads (list, validate, verify, browser, doctor, plugin...) * fix(ci): restore package-lock.json from main (unrelated lockfile churn)
2.9 KiB
AI Workflow
OpenCLI is designed for AI agents writing adapters. The workflow is built on a small set of browser primitives plus a skill that teaches the end-to-end loop.
The Loop
From a new site URL to a passing opencli browser verify — one skill, one set of primitives:
# 1. Pick up the skill (Claude Code)
# skills/opencli-adapter-author/SKILL.md
# 2. Reconnaissance
opencli browser open https://example.com
opencli browser wait time 3
opencli browser network # inspect XHR / fetch calls
opencli browser state # extract __INITIAL_STATE__ / __NEXT_DATA__
# 3. Scaffold + verify
opencli browser init <site>/<name>
opencli browser verify <site>/<name>
The skill opencli-adapter-author walks through: coverage self-test → site recon → API discovery → field decoding → output design → adapter coding → verify → write-back to site memory.
See skills/opencli-adapter-author/SKILL.md.
Primitives
| Command | Purpose |
|---|---|
opencli doctor |
Sanity check: bridge, Chrome, signals |
opencli browser open <url> |
Open a tab in the Chrome session |
opencli browser network |
List recent XHR / fetch calls |
opencli browser state |
Page state: URL, title, interactive elements |
opencli browser eval '<expr>' |
Evaluate JS in the page context (cookies + origin honored) |
opencli browser init <site>/<name> |
Scaffold ~/.opencli/clis/<site>/<name>.js |
opencli browser verify <site>/<name> |
Run the adapter and print first rows |
No explore / synthesize / generate / cascade command. The skill drives the loop — the primitives are small and composable.
Site Memory
Every site accumulates knowledge at ~/.opencli/sites/<site>/ (endpoints, field decode map, notes, response fixtures). The adapter-author skill reads memory on Step 2 and writes back on Step 12 — see skills/opencli-adapter-author/references/site-memory.md for the schema.
In-repo seeds for well-known sites live at skills/opencli-adapter-author/references/site-memory/<site>.md (eastmoney / xueqiu / bilibili / tonghuashun already covered).
Authentication Strategies
Adapters declare one of:
- PUBLIC — direct fetch, no credentials
- COOKIE — reuse Chrome session cookies (
browser: true+credentials: 'include') - HEADER — inject a custom header (bearer / csrf / signed token)
- INTERCEPT — let the page make the request; capture the response
Pick per the coverage-matrix.md and api-discovery.md references inside the skill.
When Something Breaks
- Verify failure → run
opencli doctor, then consultskills/opencli-autofix/SKILL.md - Field values wrong → jump back to
skills/opencli-adapter-author/references/field-decode-playbook.md - Endpoint returns 401/403 →
api-discovery.md§4 (token) / §5 (intercept)