agentleFS
BI-BlitzerBI-Blitzer/tandem-tokens/AGENTS.md

tandem-tokens

AGENTS.md1 starsChanged 4 months agoMITView on GitHub

This project is ollamaq / tandem-tokens: a local Ollama harness that does map-phase work so premium agents (Codex, Claude, Grok, …) spend tokens only on orchestration and reduce/clean. Full policy skill: skills/hybrid/SKILL.md Claude-oriented copy: CLAUDE.md + .claude/skills/{offload,harvest,hybrid}/ MCP setup by client: clients/README.md · MCP.md 1. Worker output stays on disk under results/<job_id>/. 2. Read only results/<job_id>/digest.md (or MCP queue_digest / hybrid_digest) — not bulk merged.jsonl unless you are deliberately reducing a small residue. 3. Never paste bulk worker output into…

  • Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.

    Of the AGENTS.md files we read, the 32,671 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.

  • In this repository5 other files it gives its agents.

The file

We split this AGENTS.md into the rules it gives an agent

3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read only results/<job_id>/digest.md (or MCP queue_digest / hybrid_digest) — not bulk merged.jsonl unless you are deliberately reducing a …

    line 14
  2. Distinctive:

    Never paste bulk worker output into the chat.

    line 17
  3. Very specific:

    When done → digest only → reduce residue → final artifact.

    line 68

≈ “Common” is an estimate: at least 1 in 5 of the 41,581 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

AGENTS.md — hybrid orchestrator (Codex, Cursor, Grok CLI, others)

This project is ollamaq / tandem-tokens: a local Ollama harness that does map-phase work so premium agents (Codex, Claude, Grok, …) spend tokens only on orchestration and reduce/clean.

Full policy skill: skills/hybrid/SKILL.md
Claude-oriented copy: CLAUDE.md + .claude/skills/{offload,harvest,hybrid}/
MCP setup by client: clients/README.md · MCP.md

Contract (do not break)

  1. Worker output stays on disk under results/<job_id>/.
  2. Read only results/<job_id>/digest.md (or MCP queue_digest / hybrid_digest) — not bulk merged.jsonl unless you are deliberately reducing a small residue.
  3. Never paste bulk worker output into the chat.

Offload bar

Offload when ≥ ~20 similar independent items and the work is mechanical (map), generate-and-test (gen_test), or staged (pipeline). Otherwise do it inline — setup cost exceeds savings.

Commands (CLI — works without MCP)

ollamaq doctor
ollamaq recommend
ollamaq submit <job.json> --job-id <name>
ollamaq run <job.json> --wait          # small batches
ollamaq status
ollamaq digest <job_id>
ollamaq score                          # tokens + $ avoided + GPU Wh
ollamaq daemon                         # one per machine

MCP tools (Codex / Cursor / generic)

Server: python -m ollamaq.mcp_server
Config snippet: clients/codex.config.toml or clients/cursor.mcp.json.

ToolUse
queue_submitBulk offload
queue_statusPoll done/failed
queue_digestHarvest summary only
ollama_quickTiny one-shot (capped; not for bulk)
recommend_modelsVRAM fit
hybrid_scoreSavings scoreboard
hybrid_energyLive GPU watts
hybrid_policyThis policy as tool text

Grok clients may use python -m ollamaq.mcp_grok for hybrid_* aliases of the same tools.

Routing

model: "auto" + task_tag in the job spec → config/models.yaml. Default workhorse is the local 14B-class model. Optional residual escalate: retry.escalate_to: "auto" or cloud:grok (needs key + cloud.grok.enabled).

Workflow

  1. Confirm offload-eligible → write jobs/<name>.job.json.
  2. ollamaq submit / MCP queue_submit (daemon running).
  3. When done → digest only → reduce residue → final artifact.
  4. Report ollamaq score savings (tokens, $, Wh when metered).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.