sutro-context
Skill16 starsChanged 6 months agoUnlicenseView on GitHub
Use before any research task, experiment, or PR review. Loads current project state from DISCOVERIES.md, open questions, and recent Telegram discussion.
In this repository9 other files it gives its agents.
- AGENTS.mdAGENTS.md
- .codex/AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- anti-slop-guideSkill
- info-defragSkill
- prepare-meetingSkill
- run-experimentSkill
- sutro-syncSkill
- weekly-catchupSkill
The file
We split this skill into the rules it gives an agent
7 rules. About 0 are common, 3 are distinctive, and 4 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
4
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 12
Read these in order. Stop and report if anything is unexpected.
- Distinctive: line 21
Read the last 10 messages from priority Telegram topics:
- Distinctive: line 36
Check GitHub for open work:
- Very specific: line 56
Check research/search_space.yaml for allowed parameter ranges
- Very specific: line 57
Check research/questions.yaml for the dependency graph of open questions
- Very specific: line 58
Run checks/env_check.py to verify environment
- Very specific: line 59
Run checks/baseline_check.py if baselines may differ on your machine
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill 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.
Files
Sutro Research Context
Load this before doing any research work, running experiments, or reviewing PRs.
Step 1: Read core files
Read these in order. Stop and report if anything is unexpected.
- CLAUDE.md -- project context, current best methods, constraints
- DISCOVERIES.md -- what's proven, what failed, open questions (bottom of file)
- AGENT.md -- machine-executable experiment loop (if running autonomous)
- LAB.md -- experiment protocol, rules (especially rule #9: metric isolation)
Step 2: Check recent activity
Read the last 10 messages from priority Telegram topics:
import json
for f in ['chat-yad.json', 'chat-yaroslav.json', 'challenge-1-sparse-parity.json']:
path = f'src/sparse_parity/telegram_sync/{f}'
try:
msgs = json.load(open(path))
print(f'\n=== {f} (last 3) ===')
for m in msgs[:3]:
print(f" [{m['date'][:10]}] {m['sender']}: {m['text'][:150]}")
except FileNotFoundError:
print(f'{f} not found -- run: bun run sync_telegram.ts')Check GitHub for open work:
gh pr list --repo cybertronai/SutroYaro --state open
gh issue list --repo cybertronai/SutroYaro --state openStep 3: Know the current state
| Fact | Value |
| Best method | GF(2) Gaussian elimination, 509us, ARD ~500 |
| Best energy proxy | DMC (Data Movement Complexity, Ding et al.) |
| Experiments done | 33+ (see research/log.jsonl) |
| Open questions | Bottom of DISCOVERIES.md (Q7, Q11-Q13 still open) |
| Next milestone | Energy-efficient nanoGPT training ("final exam") |
| Meeting cadence | Mondays 18:00 at South Park Commons |
Step 4: Before writing code
- Check
research/search_space.yamlfor allowed parameter ranges - Check
research/questions.yamlfor the dependency graph of open questions - Run
checks/env_check.pyto verify environment - Run
checks/baseline_check.pyif baselines may differ on your machine
For contributors using other tools
This context applies regardless of which AI tool you use. The key files are plain markdown and YAML. Read them before starting work. The experiment protocol in LAB.md and AGENT.md defines the loop: hypothesis, code, run, measure, record.
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
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.

