skills
CLAUDE.md0 starsChanged 2 months agoMITView on GitHub
Public Agent Skills that teach coding agents to build on the Talkif API. This repo is public: the same hygiene as Talkif-ai/docs — no internal service names, hosts, infrastructure, PR numbers or people; brand is Talkif / Talkif-ai only. One skills/<name>/SKILL.md per job an agent might be asked to do ("place a call", "verify a webhook"). Frontmatter name + description (the description is what an agent matches on — say when to use it). Body order: concepts → exact request…
- Reads credentials
In this repository22 other files it gives its agents.
- api-conventionsSkill
- choose-models-and-voicesSkill
- create-campaignSkill
- create-flow-functionSkill
- create-flowSkill
- create-scheduleSkill
- do-not-callSkill
- handle-inbound-callsSkill
- make-callSkill
- manage-contactsSkill
- manage-phone-numbersSkill
- quickstartSkill
- read-billingSkill
- read-call-resultsSkill
- realtime-eventsSkill
- setup-api-keySkill
- test-a-flowSkill
- troubleshoot-callsSkill
- update-flowSkill
- verify-webhooksSkill
- web-callsSkill
- write-promptsSkill
The file
We split this CLAUDE.md into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 18
Never write a method name or field from memory.
≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
CLAUDE.md — Talkif-ai/skills
Public Agent Skills that teach coding agents to build on the Talkif API. This repo is public: the same hygiene as Talkif-ai/docs — no internal service names, hosts, infrastructure, PR numbers or people; brand is Talkif / Talkif-ai only.
What a skill is here
One skills/<name>/SKILL.md per job an agent might be asked to do ("place a call", "verify a webhook"). Frontmatter name + description (the description is what an agent matches on — say when to use it). Body order: concepts → exact request (JSON) → TypeScript SDK → Python SDK → curl where useful → gotchas → docs link. 80–150 lines. No dashboard click-instructions: if a feature has no public endpoint (provider setup, Meta Lead Ads, WhatsApp, top-ups, API-key creation) it's a one-line pointer to the docs, not a skill.
Sources of truth — every fact must come from one of these
| Fact | Source |
| Endpoints, fields, enums, query params | https://api.talkif.ai/openapi.json (mirrored as fern/openapi.json in Talkif-ai/docs) |
| SDK method names | the generated clients: Talkif-ai/talkif-typescript src/api/resources/*/client/Client.ts; Python is the snake_case of the same names |
| Behaviour, limits, defaults | docs.talkif.ai guides (they are verified against the platform); when the spec text and the guide disagree, the guide wins and the spec defect is filed upstream |
| Browser SDK | Talkif-ai/webrtc-js README |
Never write a method name or field from memory. Cross-check every METHOD /path against the spec before committing (a scripted check: extract all `GET /…` strings and assert they exist in openapi.json, allowing /api/v1 and {param} differences).
Recheck rule
Skills drift silently. Recheck all of them when any of these change: the public spec (openapi.json), an SDK release, docs.yml slugs (every Docs: link must resolve), the browser SDK API. Search skills/ for the changed endpoint/method/slug and update every hit in the same PR; add a changelog line in Talkif-ai/docs if the change is customer-visible.
Known "not yet" facts baked into skills (flip when they ship): no client idempotency key on POST /calls; call transfer not available; the call list is GET /calls, calls.listCalls in SDK 0.3.0 (getCallHistory / getActiveCalls were removed in 0.2.0), and it returns one page, not a pager.
Conventions
- Skill names are verbs or verb-phrases in kebab-case (
make-call,verify-webhooks);quickstartandapi-conventionsare the two exceptions and the two entry points. - Code samples must run as written with
TALKIF_API_KEYset; placeholders in<angle-brackets>. - Escape
\$only where a Markdown renderer could treat$…$as math; keep prose plain. - Commits:
feat: <skill>,fix: <skill>: <what>,chore:. No attribution trailers. - The README table lists every skill; update it when adding one.
Checks before committing
grep -rniE "backend-rs|catflow|kamailio|nats|redis|postgres|consul|kubernetes|argo|terraform|pipecat|webhook-executor" skills README.md # must be emptyplus the endpoint cross-check above.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.