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 commands/adilkalam/orca/researchgit clone --depth 1 https://github.com/adilkalam/orcaWhat 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 | $0.00014 | $0.02995 |
| Opus 5 | $0.00007 | $0.01497 |
| Sonnet 5 | $0.00003 | $0.00599 |
| Haiku 4.5 | $0.00001 | $0.00299 |
Grade A, and why
research scanned grade A with 1 finding against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured yesterday.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
if curl -s -m 3 http://localhost:11235/ >/dev/null 2>&1; then How it starts
The opening of the file, as written. The whole thing — 364 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/research - Research Lane Orchestrator
YOUR ROLE: DIRECT ORCHESTRATION
You are the orchestrator, running in the main thread. You spawn all subagents
directly via the Agent tool — there is no lead agent. Claude Code forbids
subagents from spawning subagents, so all delegation happens here, single-level,
one at a time. See docs/reference/flatten-orchestration-pattern.md.
Your subagents:
research-web-search-subagent- WebSearch + Crawl4AI for web queriesresearch-site-crawler-subagent- Deep crawling of specific domainsresearch-answer-writer- Standard research reportsresearch-deep-writer- Long-form deep researchresearch-citation-gate- Citation verification (checks claims againstsources/)research-fact-checker- Re-verification gate (always-on under--deep, opt-in via--verify)research-consistency-gate- Consistency checks
Phase 0: Create Research Folder (MANDATORY FIRST STEP)
BEFORE doing anything else, create a dated research folder:
# Generate folder name: YYYY-MM-DD-Topic-Slug
# Example: 2025-12-25-Technical-Trading-Patterns
RESEARCH_DIR=".orca/research/$(date +%Y-%m-%d)-<Topic-Slug>"
mkdir -p "$RESEARCH_DIR/evidence" "$RESEARCH_DIR/sources"
Naming rules for Topic-Slug:
- Use 2-4 words from the research question
- Title case with hyphens
- Examples:
LLM-Reasoning-Techniques,Chart-Pattern-Detection,React-State-Management
All research output goes into this folder:
$RESEARCH_DIR/evidence/- Evidence Notes from subagents$RESEARCH_DIR/sources/- RETAINED raw fetched/crawled pages (primary evidence — never deleted)$RESEARCH_DIR/synthesis.md- Evidence synthesis$RESEARCH_DIR/report.md- Final research report$RESEARCH_DIR/citation-audit.md- Citation-gate audit output$RESEARCH_DIR/sources/temp/- Scratch for in-flight crawl data ONLY; partial files may be cleaned, retained raw pages may NOT
CRITICAL: Pass the $RESEARCH_DIR path to ALL subagents in their prompts.
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.
- yesterday First seen · 364 lines · 14 tokens per session scan A 8bf0fa233250
research is a command published in the GitHub repository adilkalam/orca (2 stars, last pushed 1mo ago), licensed MIT. It adds 14 tokens to every session and 2,995 once invoked, about $0.0001 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 commands, from other repositories
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
specify
Create or update the feature specification from a natural language feature description.
analyze
Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md after task generation.
converge
Assess the current codebase against the feature's spec, plan, and tasks, then append any remaining unbuilt work as new tasks to tasks.md so implement can complete it.
implement
Execute the implementation plan by processing and executing all tasks defined in tasks.md.