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 buildinternet/releases --skill local-ingestgit clone --depth 1 https://github.com/buildinternet/releasesWrote 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/buildinternet/releases/local-ingest)<a href="https://agentmods.dev/skills/buildinternet/releases/local-ingest"><img src="https://agentmods.dev/badge/skills/buildinternet/releases/local-ingest.svg" alt="Measured on agentmods" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 3 findings, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Prompt Injection · line 48 Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.Fix: Remove the large whitespace padding (blank-line blocks or long space runs) and review any content hidden below or to the right of it. Keep skill files compact and reviewable so no instructions can be
- medium Data Exfiltration · line 104 Data is being sent to an external URL. This could be legitimate telemetry or data exfiltration. Manual review is recommended.Fix: Verify the destination URL is trusted and necessary. Remove or replace with documented APIs. Ensure no secrets, tokens, or PII are transmitted.
- medium Prompt Injection · line 168 Large whitespace padding was detected (a block of blank lines or a long run of spaces). This can push injected instructions below or to the right of the visible area so a human reviewer never sees them while the agent still reads them. Manual review of the hidden content is recommended.Fix: Remove the large whitespace padding (blank-line blocks or long space runs) and review any content hidden below or to the right of it. Keep skill files compact and reviewable so no instructions can be
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 | $0.00100 | $0.04380 |
| Opus 5 | $0.00050 | $0.02190 |
| Sonnet 5 | $0.00020 | $0.00876 |
| Haiku 4.5 | $0.00010 | $0.00438 |
Grade A, and why
local-ingest scanned grade A with 1 finding 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 8d 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
- **Client-rendered index** → the index is a JS shell whose cards load via a client-side fetch, so `curl`/non-JS scrape returns **zero** entries. Enumerate by rendering: drive the index with Claude in Chrome (or fan out How it starts
The opening of the file, as written. The whole thing — 201 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Local Ingest
Onboard a changelog without dispatching the remote fetch path. The remote path (releases admin source fetch → POST /v1/workflows/update → the API worker's DeterministicUpdateWorkflow, #1946) runs the body→records extraction server-side (incremental Haiku or the tool-loop) and bills that inference — even when it writes zero releases. A local Claude Code agent is already a capable model with web-fetch + file tools, so it can do the per-page extraction inline (folded into the session you're already in) and only needs a deterministic write path.
Core principle: the agent is the extractor. You fetch, you parse, you structure the records, and you POST them to the batch endpoint, which does a plain idempotent upsert with no AI on insert. The remote extraction loop never runs.
Local Claude Code only. This skill uses the Agent tool for sub-agent fan-out and assumes a persistent local filesystem and the CLI's RELEASES_API_* env. It sits alongside seeding-playbooks as a local-only operator skill — it is not deployed to the managed-agent fleet.
Cost contract
State this explicitly to the operator before a large run, and honor it:
- Spends: your agent's own tokens for fetch + extract, plus any sub-agents you fan out (your choice of model —
sonnetfor quality,haikufor bulk). - Does NOT spend: no server-side extraction (the update workflow's incremental-Haiku / tool-loop passes), no remote
web_fetchextraction.POST /v1/workflows/updateis never called. - The batch endpoint runs no AI on insert — no summarization, no marketing classifier, no feed enrichment. It does an
insertplus a fire-and-forget vector embed (waitUntil). The AI fields (title_generated/title_short/summary) are intentionally left empty; see Manual enrichment below for the deliberate follow-up paths.
The negative signal that confirms no remote run happened: in Axiom releases-cloudflare-logs there is no POST .../fetch?sessionId=det-… and no extract-deps-worker event for the source.
What ships with it
1 file 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.
- 8d ago First seen · 201 lines · 100 tokens per session scan A c3c69a12ee6f
local-ingest is a skill published in the GitHub repository buildinternet/releases (6 stars, last pushed 4d ago), licensed Apache-2.0. It adds 100 tokens to every session and 4,380 once invoked, about $0.0005 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other skills, from other repositories
xquik-social-data
Use Xquik for X and Twitter social data workflows through its public API, SDKs, MCP server, webhooks, and installable agent skill.
releases-cli
Use the releases CLI to search, browse, and read the Releases.sh changelog registry from the terminal — the keyless, agent-friendly peer of the Releases MCP. Activate when the user mentions "releases CLI", runs a releases command, asks how to install it, or wants to look up releases, sources, orgs, collections, or…
releases-mcp
Use when the user asks about recent releases, changelogs, what's new in a library, breaking changes, version updates, or wants to compare products. Activates for questions like "what changed in Next.js 15?", "latest Tailwind releases", "compare Bun vs Deno releases".
analyzing-releases
Analyze release trends across multiple companies to produce competitive intelligence. Use when asked to compare companies, analyze a market segment, identify industry trends, forecast upcoming releases, or answer questions like "what is X shipping lately" or "how does X compare to Y." Also triggers on requests for…
graphql-expert
GraphQL expert for schema design, resolvers, subscriptions, and performance optimization.
openapi-expert
OpenAPI/Swagger expert for API specification design, validation, and code generation.