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.
git clone --depth 1 https://github.com/luutuankiet/gsd-liteWrote 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/luutuankiet/gsd-lite/gsd-lite)<a href="https://agentmods.dev/agents/luutuankiet/gsd-lite/gsd-lite"><img src="https://agentmods.dev/badge/agents/luutuankiet/gsd-lite/gsd-lite/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/luutuankiet/gsd-lite/gsd-lite"><img src="https://agentmods.dev/badge/agents/luutuankiet/gsd-lite/gsd-lite.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00022 | $0.03410 |
| Opus 5.5 | $0.00009 | $0.01364 |
| Sonnet 5.5 | $0.00004 | $0.00682 |
| Haiku 4.5 | $0.00002 | $0.00341 |
Grade A, and why
gsd-lite 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 1mo 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.
| A) Token mismatch | High | Manual curl test | CONFIRMED | How it starts
The opening of the file, as written. The whole thing — 348 lines — stays where its author put it; the contents beside it link to each section on GitHub.
GSD-Lite Protocol
1. Safety Protocol (CRITICAL)
NEVER overwrite existing artifacts with templates.
Before writing to WORK.md or INBOX.md:
- Check existence: read
gsd-lite/directory - Read first: If file exists, read it to understand current state
- Append/Update: Only add new information or update specific fields
- Preserve: Keep all existing history, loops, and decisions
2. Universal Onboarding (CRITICAL)
MUST be completed on EVERY first turn — even if user gives direct instruction.
If user says "look at LOG-071" on turn 1, respond: "I'll get to LOG-071 right after I review the project context."
Boot sequence — bounded reads, in this order:
gsd-lite/PROJECT.md— project vision. Whole file; it is small by design and its File Index restates the contract below.gsd-lite/ARCHITECTURE.md— technical landscape, zone A only.rg -n '^## Version Archive' gsd-lite/ARCHITECTURE.md→ line N, then read withlimit: N.gsd-lite/WORK.md— current state, sections 1 and 2 only.rg -n '^## 2b\.' gsd-lite/WORK.md→ line N, then read withlimit: N.
Never read ARCHITECTURE.md or WORK.md without a bound. Both are append-only and grow without limit. On a mature project the architecture file reaches tens of thousands of words and the work log thousands of lines — an unbounded read of either spends most of a context window before any work starts. This is the single most expensive mistake available on turn 1.
If a boundary rg returns no match, the file is either a fresh template or predates the archive convention. Read it whole this once, then add the missing boundary heading before it grows.
Section 3 of WORK.md is index-then-read, never scan — including when the user opens with "look at LOG-071". rg -n '^### \[LOG-' gsd-lite/WORK.md gives the index; read only that entry's line range.
Heading grep hygiene: grep these artifacts with ^#{2,4} , never ^#{1,4} . Fenced code blocks contain shell comments starting with # , and a level-1 grep reports those as headings — producing an index that looks trustworthy and is not.
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.
- 1mo ago First seen · 348 lines · 22 tokens per session scan A 5aa78319e7f3
gsd-lite is an agent published in the GitHub repository luutuankiet/gsd-lite (3 stars, last pushed 2mo ago), licensed MIT. It adds 22 tokens to every session and 3,410 once invoked, about $0.0001 per session on Opus 5.5. 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 agents, from other repositories
CocoPull
Lossless context distillation agent. Produces dense pull artifacts that preserve decision-bearing facts from large files.
onboard
Analyze brownfield codebase and create initial continuity ledger.
chronicler
Session analysis, precedent lookup, and learning extraction.
context-query-agent
Query the artifact index for precedent and guidance.
source-ingester
An agent for turning one outside source, such as an article, book, video, or lecture note, into linked notes in an Obsidian knowledge base.
memory-optimizer
Recommends the cheapest direct-read, Markdown, graph build, update, or query path for a corpus while keeping graphify as the primary long-lived memory engine.