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 skills/harnessprotocol/harness-kit/refresh-researchnpx skills add harnessprotocol/harness-kit --skill refresh-researchgit clone --depth 1 https://github.com/harnessprotocol/harness-kitWhat 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 | $0.00052 | $0.03752 |
| Opus 5 | $0.00026 | $0.01876 |
| Sonnet 5 | $0.00010 | $0.00750 |
| Haiku 4.5 | $0.00005 | $0.00375 |
Grade A, and why
refresh-research 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 — 403 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Research Refresh Protocol
Overview
Re-fetch live research sources (repos, docs), detect meaningful changes since initial capture, and update synthesis files. Raw sources are versioned (never overwritten), syntheses are surgically updated, and INDEX.md is rebuilt.
Core principles:
- Only refreshable sources are fetched.
repoanddocssource types. Blogs, papers, videos, and internal docs are static — skip them. - Raw sources are versioned, never overwritten. New fetch → new file with today's date.
- Syntheses are updated, not replaced. Preserve analysis and structure; update facts.
When to Use
User types /refresh-research followed by:
| Invocation | Behavior |
|---|---|
| (empty) | Interactive — show stale entries, let user pick which to refresh |
research/agent-memory/cognee-architecture.md |
Refresh a specific synthesis file |
agent-memory |
All refreshable entries in that category |
--all |
All refreshable entries across all categories |
--stale 14 |
Entries not checked in 14+ days (default threshold: 30) |
--backfill |
One-time: add source_type to entries missing it |
--audit |
Dry-run: staleness report, no modifications |
Flags can be combined: /refresh-research agent-memory --stale 7
Frontmatter Fields
Three optional fields extend the existing synthesis frontmatter:
---
tags: [github, memory, open-source]
date: 2026-02-12 # original research date (unchanged meaning)
source: https://docs.cognee.ai/
source_type: docs # NEW — repo | docs | blog | paper | internal | video
last_checked: 2026-03-16 # NEW — when last verified against source
refresh_status: changed # NEW — changed | unchanged | dead-link | archived
---
source_typegates refresh eligibility — onlyrepoanddocsare refreshablelast_checkedis the staleness clock (separate fromdatewhich means "first captured")refresh_statusis informational only (doesn't gate behavior)
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.
- 2d ago First seen · 403 lines · 52 tokens per session scan A 14816c04a074
refresh-research is a skill published in the GitHub repository harnessprotocol/harness-kit (10 stars, last pushed 3d ago), licensed Apache-2.0. It adds 52 tokens to every session and 3,752 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 skills, from other repositories
studio
Architecture Studio control plane — initialize or inspect a studio workspace, create and register projects, or route an architecture/AEC task to the right agent or skill. Use when the user runs /as:studio, asks to set up or open their studio, manage its projects, or describes a task without naming a skill.
epd-to-spec
Write CSI specification language for EPD submittals and sourced GWP limits. Use to add embodied-carbon or EPD requirements to specs; not to parse, find, or compare EPDs.
occupancy-calculator
Calculate code occupant loads by area with gross/net factors and jurisdiction checks. Use for "how many people can this space hold," IBC Table 1004.5, egress inputs, or occupancy-load reports; not for workplace headcount planning.
zoning-analysis-nyc
Analyze NYC lot zoning, FAR, height, setbacks, uses, and buildable envelope from PLUTO and the Zoning Resolution. Use for "what can I build"; use nyc-bsa for relief and zoning-envelope for 3D visualization.
spec-writer
Turn a material or product list into CSI MasterFormat outline specifications with three-part sections and review flags. Use to write specs; use epd-to-spec for embodied-carbon requirements.
workplace-programmer
Build office space programs through guided workplace strategy: area splits, room schedules, and planned seat counts. Use to program or size a workplace; not for code occupant-load calculations.