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 instructions/truefoundry/trueforge/agents-mdgit clone --depth 1 https://github.com/truefoundry/trueforgeWhat 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 | $0.01070 | $0.01070 |
| Opus 5 | $0.00535 | $0.00535 |
| Sonnet 5 | $0.00214 | $0.00214 |
| Haiku 4.5 | $0.00107 | $0.00107 |
Grade A, and why
trueforge AGENTS.md 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 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.
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 — 23 lines — stays where its author put it; the contents beside it link to each section on GitHub.
- Workspace tasks MUST use
package.jsonscripts; add a script when a missing workflow is repeatable like the existing commands, not ad hoc commands. - CI package path filters, matrix package ids, and root scripts
test:*in.github/workflows/ci.ymland rootpackage.jsonMUST stay synchronized when a workspace package is added, renamed, or moved; thestorefilter sync rule lives inpackages/trueforge-core/src/agent-session/store/AGENTS.md. - Release wiring MUST keep dist-free host development,
pnpm smoke, and packed CJS/ESM consumers of@truefoundry/trueforge-coreworking without changes. - PRs that change published-package code (
packages/trueforge-core,packages/trueforge,packages/trueforge-ui,packages/trueforge-sdk) orpackages/frontend(ships inside@truefoundry/trueforge) MUST include a new.changeset/*.mdfile (pnpm changeset). Docs, CI/workflows, charts, and docker-compose changes do not. SDK regeneration already adds@truefoundry/trueforge-sdkviapnpm changeset:sdk-regen. - Shared Postgres/Redis settings in
docker-compose.ymlanddocker-compose.dev.yml(image versions, health checks,env_file) MUST stay synchronized; intentional differences (app services, data paths, projectname, host ports, in-networkPOSTGRES_HOST/REDIS_URL) MUST stay explicit.packages/trueforge/.envis the host-dev + secrets source;docker-compose.ymlmay read it but MUST override container connectivity so host-dev localhost values are not used inside the smoke-test stack. - Changes to types or schemas MUST keep
packages/trueforge-core,packages/frontend,packages/trueforge, andpatchessynchronized; they MUST NOT update only one affected layer. - TypeScript code MUST NOT use assertion escapes such as
as T,as unknown as T, non-null!, oras neverto silence type errors; implementations MUST use sound contracts, guards, or corrected types. - When catching an error and throwing another, the new error MUST set
{ cause: caught }so the original failure is preserved for logs and debugging. - Every type, schema, helper, and contract MUST have one canonical owner; code MUST NOT introduce duplicate definitions or forwarding shims that hide ownership.
- AgentUIServer / server-port types (
AgentChatServer,AgentBuilderServer, catalog ports, session/turn DTOs, stream events) MUST be defined only in@truefoundry/assistant-ui-runtime(src/server/types.ts/events.ts).@truefoundry/trueforge-uiMUST re-export them fromsrc/server/types.ts(pass-through aliases only) and MUST NOT add hand-written parallel definitions of those ports/DTOs. - Runtime types backed by Zod schemas MUST be derived with named
z.infer<typeof Schema>aliases; code MUST NOT duplicate those schemas as hand-written interfaces or indirect utility-type chains. - Modules MUST use static
importandimport type; they MUST NOT userequire(),require.resolve(), or lint suppressions to bypass import checks. - A change that makes code unused MUST remove that dead code in the same change; it MUST NOT leave stale exports, files, documentation, duplicates, or “just in case” shims.
- Comments MUST explain intent, trade-offs, or constraints and remain concise; they MUST NOT restate the code or include issue-tracker IDs.
- Every nested
AGENTS.mdMUST have a siblingCLAUDE.mdcontaining only@AGENTS.md, so Cursor and Claude Code load the same scoped rules. - OpenAPI specs under
.github/fern/openapi/openapi.jsonanddocs/openapi.jsonare autogenerated in CI and MUST NOT be edited manually (both copies MUST stay identical) - Code under
packages/trueforge-sdkis autogenerated in CI and MUST NOT be edited manually - Tests MUST live under a package-top-level
testortestsdirectory that mirrors thesrclayout; they MUST NOT be inlined undersrc. - Functions with more than one parameter of the same type MUST take a single options object (e.g.
f({ a, b }: { a: string; b: string })); they MUST NOT use multiple positional parameters of that repeated type (e.g.f(a: string, b: string)). - Zod unions MUST use
z.discriminatedUnionwhenever every member carries a shared literal discriminator field;z.unionis only permitted when no such common discriminator exists. - HTTP/OpenAPI wire shapes (path params, query params, request/response JSON fields) and database identifiers (table/column names, persisted jsonb document keys) MUST use
snake_case. - Server code MUST NOT read environment variables via
process.envdirectly; all env reads MUST go throughpackages/trueforge/src/config.ts, unless there is a documented special requirement (for example bootstrap before config is loaded).
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 · 23 lines · 1,070 tokens per session scan A e685cd2cffae
trueforge AGENTS.md is an instructions file published in the GitHub repository truefoundry/trueforge (4,935 stars, last pushed 2d ago), licensed MIT. It adds 1,070 tokens to every session, about $0.0053 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 2026-08-30.
Other instructions, from other repositories
learn-harness-engineering CLAUDE.md
Instructions for walkinglabs/learn-harness-engineering, covering claude.md, project overview, commands, documentation site and run lecture code examples.
relay AGENTS.md
Instructions for AgentWorkforce/relay, covering git workflow rules, never push directly to main, correct workflow, ... do work .. and stop here - let user merge.
LeAgent AGENTS.md
Instructions for vixues/LeAgent, covering leagent development guidelines, project overview, architecture, execution topology (one kernel, many ingresses) and surface map (where things live).
grix AGENTS.md
Instructions for askie/grix, covering grix agent guide, repository boundaries, required workflows, shared agent configuration and cross-component contracts.
opencode-plusplus AGENTS.md
Instructions for whut09/opencode-plusplus, covering agent guide, generated agent guide, must-read rules, default workflow and project entrypoints.
sprawling CLAUDE.md
Instructions for 2youg1/sprawling: See AGENTS.md. It is the whole instruction set for this repository — the loop, what to read first, the five steps of one change, the rules a machine holds, and the language and commit conventions.