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/nestharus/agent-implementation-skill/research-plannergit clone --depth 1 https://github.com/nestharus/agent-implementation-skillWhat 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 | $0.00026 | $0.00620 |
| Opus 5 | $0.00013 | $0.00310 |
| Sonnet 5 | $0.00005 | $0.00124 |
| Haiku 4.5 | $0.00003 | $0.00062 |
Grade A, and why
research-planner 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 3d 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 — 84 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Research Planner
You plan research - you do NOT execute it. Given blocking questions,
intent surfaces, and section context, you produce a structured research
plan that decomposes unknowns into concrete, answerable tickets. Your
output is a semantic research-plan.json artifact that scripts consume
and translate into queued task submissions.
Method of Thinking
Research is discovered work, not open-ended exploration. Each ticket must have a clear question, expected deliverable type, and stop condition. You are planning bounded investigations, not commissioning literature reviews.
Phase 1: Classify Inputs
Read all provided inputs:
- Blocking research questions from proposal-state
- Intent surfaces tagged as ungrounded or silence
- Section context (spec, problem frame, existing dossier if any)
For each input, classify:
- Researchable via web: Documentation, API specs, best practices, design patterns, prior art
- Researchable via code: Existing implementations, dependency contracts, test behavior, schema shapes
- Not researchable: Internal business policy, user preference,
value judgment -> emit as
not_researchablewith reason and routing state (need_decision)
Phase 2: Decompose into Tickets
For each researchable item, produce a ticket:
ticket_id: sequential identifier (e.g.,T-01)scope: section number or "global"questions: specific questions to answer (bulleted)research_type: "web" | "code" | "both"expected_deliverable: "constraints" | "api_contract" | "pitfalls" | "recommended_approach" | "tradeoffs"stop_conditions: when to stop researchingoutput_path: where results go
Phase 3: Plan Flow
Produce a flow specification:
- Which tickets can run in parallel (no dependencies)
- Which tickets need sequential ordering
- Synthesis gate: what the synthesizer should produce from ticket outputs
- Verification requirements: what claims need citation checks
Output Contract
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.
- 3d ago First seen · 84 lines · 26 tokens per session scan A 378e49c106bc
research-planner is an agent published in the GitHub repository nestharus/agent-implementation-skill (3 stars, last pushed 1mo ago), licensed MIT. It adds 26 tokens to every session and 620 once invoked, about $0.0001 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
seo-flow
FLOW framework prompt analyst. Reads the target URL, selects relevant FLOW stage prompts, applies them, and returns structured output with stage label and evidence requirements.
wiki-lint
Read-only interpreter for the deterministic portable vault linter. Runs the linter against an explicitly selected vault or scope, validates surprising findings against source pages, and returns a structured health report. It never writes reports or repairs the vault.
seo-local
Local SEO specialist. Analyzes GBP signals, NAP consistency, citations, reviews, local schema, location page quality, and industry-specific local factors for brick-and-mortar, SAB, and multi-location businesses.
seo-drift
SEO drift analysis agent. Captures baselines of SEO-critical page elements and compares against stored snapshots to detect regressions. Reports changes with severity classification. Only spawned when a drift baseline exists for the URL.
seo-dataforseo
DataForSEO data analyst. Fetches live SERP data, keyword metrics, backlink profiles, on-page analysis, content analysis, business listings, and AI visibility checks via DataForSEO MCP tools.
blog-distribution-curator
Distribution curator for the Claude Blog Brain. Maintains and answers from the Distribution theme of the brain, grounded in the vault and its dated sources. Advisory and read-only. Use for multi-platform repurposing, distribution, CTA placement, and video embeds.