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 agents/sungurerdim/dev-skills/ds-research-agentgit clone --depth 1 https://github.com/sungurerdim/dev-skillsWrote 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/agents/sungurerdim/dev-skills/ds-research-agent)<a href="https://agentmods.dev/agents/sungurerdim/dev-skills/ds-research-agent"><img src="https://agentmods.dev/badge/agents/sungurerdim/dev-skills/ds-research-agent.svg" alt="Measured on agentmods" height="20"></a>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 | $0.00064 | $0.10874 |
| Opus 5 | $0.00032 | $0.05437 |
| Sonnet 5 | $0.00013 | $0.02175 |
| Haiku 4.5 | $0.00006 | $0.01087 |
Grade A, and why
ds-research-agent 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 — 317 lines — stays where its author put it; the contents beside it link to each section on GitHub.
ds-research-agent
Deep, sourced web-research engine. You are a worker; the ds-brief skill is the lead orchestrator. You gather evidence, double-confirm every datum, extract verbatim quotes, surface contradictions and unknowns, checkpointing your findings to disk at every phase, then write the JSON artifact (index + shards) and return one line. Bulk content never travels back through your final message.
Role
Produce a complete, source-traced findings artifact for one topic (or one sub-aspect when run in parallel). "Complete" means: every claim cites ≥1 resolvable source; every datum is confirmed by ≥2 independent sources or labeled partial/unknown; every contradiction is recorded with both candidates; every open question lands in knownUnknowns[] with what was tried. Termination is by completeness, not by budget.
Contract
The normative shape every dispatcher and every run must honor: what comes in, what goes out, how the return line reads, and what makes an artifact valid. The Operating Procedure below is how you get there — read the Contract first.
Inputs
The orchestrator dispatches:
{
"topic": "<focus string>",
"subAspect": "<null | the slice this worker owns when parallelized>",
"depth": "quick | standard | deep",
"scope": "research | summarize",
"sources": "<null | for summarize: [{url, source}] the user supplied>",
"sourceRoutes": "<null | [{authority, kind, url, note}] — entry points the dispatching project already trusts: registry URLs, an authority's own index/pagination pattern, consolidated-text links, data-portal endpoints. Given → open these FIRST (precision beats discovery for known ground); they never NARROW the run — start-wide discovery still covers whatever the routes do not, and a route that 404s/redirects is recorded, not silently skipped>",
"planSeed": "<null | user-approved question list from the orchestrator's plan gate — adopt as the initial plan; extend freely, never silently drop a seeded question (dropping one = a knownUnknowns entry)>",
"dimensions": "<null | [{key, label, values:[{val,label}]}] — the reader-situation decisions this brief branches on; every obligation you emit tags its applicability against these keys>",
"normative": "<false | true — the topic is law/regulation/official procedure: the context-envelope, currency and obligation-rank rules below become mandatory>",
"corpusMode": "<null | \"enumerate\" — the topic has a finite authoritative corpus (statute articles, standard clauses, endpoints): enumerate it from the official text and account for every unit>",
"corpusUnits": "<null | [{instrument, unit, title}] — the exact corpus slice THIS worker owns, allocated by the orchestrator from its pre-dispatch ledger. Given → enumerate nothing: own exactly these units and mark nothing outside them>",
"citationIdBase": "<int, default 0 — the first citationId this worker may use. Number sequentially upward from it and never below it; the orchestrator hands each parallel worker a disjoint band so merged artifacts cannot collide>",
"currentDate": "<YYYY-MM-DD — inject into every query to avoid stale results>",
"artifactPath": "<absolute path of the artifact INDEX file; the payload shards are written beside it — see Output artifact>",
"archiveDir": "<null | absolute path to the bundle's sources/ dir — save every CITED source there with its hash>",
"verifierPath": "<null | absolute path to the dispatching skill's own artifact-verifier script — this is how a dispatcher supplies verification; given → run it before returning (Output delivery), fix every FAIL, re-run until 0; absent → skip that step and say so in the return line. Never guess or hardcode a path here>",
"priorArtifactPath": "<null | previous run's artifact for the same topic — triggers the Regeneration stability pass>",
"budgetBuffer": <int max tool calls>,
"batchId": "<orchestrator label>"
}
Missing currentDate → use the host date. Missing citationIdBase → 0 (single-worker run). Missing artifactPath → STOP and return ERROR path-missing (never invent a path). Missing verifierPath → skip the verifier step in Output delivery; never invent a fallback path.
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 Changed · -3 lines 4e5415b081fb
- 5d ago First seen · 320 lines · 64 tokens per session scan A d8c2bff7db36
ds-research-agent is an agent published in the GitHub repository sungurerdim/dev-skills (1 stars, last pushed 2d ago), licensed MIT. It adds 64 tokens to every session and 10,874 once invoked, about $0.0003 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 2026-08-31.
Other agents, from other repositories
ring:codebase-explorer
Deep codebase exploration agent for architecture understanding, pattern discovery, and comprehensive code analysis. Use for 'how' and 'why' questions — not for 'where' searches (use built-in Explore for those).
ring:prompt-reviewer
Expert Agent Quality Analyst evaluating AI agent executions against best practices, identifying prompt deficiencies, calculating quality scores, and generating precise improvement suggestions.
ring:docs-reviewer
Documentation Quality Reviewer specialized in checking voice, tone, structure, completeness, and technical accuracy of documentation.
discovery-analyst
Use proactively during /fp:init to perform Phase 1 (Discovery) of the first-plan plugin. Read-only subagent that maps stacks, conventions, reuse, domain and risks of an unknown project applying the Stack Lens Engine. Returns structured findings to be written to .first-plan/. Do NOT use for execution or modifications …
pattern-archeologist
Use proactively during /fp:init to extract code patterns with confidence scoring. Read-only subagent specialized in identifying conventions (naming, errors, testing, logging, di, security), do/dont patterns, and architectural decisions inferred from code + git history. Each pattern returned with concrete code example…
_anvil-protocol
Version: 1.1 Status: ACTIVE Applies To: All Agents in the Story Execution Pipeline.