---
title: "Patterns: coding agents, dev tools and self-compiling workflows"
type: community
source_tier: community
tags: [ideas, patterns, coding-agents, code-review, workflows]
created: 2026-09-25
updated: 2026-09-25
confidence: medium
sources:
  - raw/x/INDEX.json
  - raw/x-repos/INDEX.json
  - raw/x/devagrawal09-2100341005690298687.md
  - raw/x-repos/jkudish__jev-mcp.md
  - raw/x-repos/valentynkit__jev-commit.md
  - raw/x-repos/doeixd__jev-pref.md
  - raw/x-repos/supercorp-ai__supercov.md
  - raw/x-repos/0x7067__claude-jev.md
  - raw/x/cyrilXBT-2101515130798297448.md
  - raw/x/MiguelriosEN-2101029313906987422.md
  - raw/x/_aj-2102061534956662818.md
jev_version: "jev-1.13.0"
summary: "P10, P11 and P37: community decision patterns for diff review and semantic lint, test-output interpretation and self-compiling workflows, each with a fit verdict."
---

# Patterns: coding agents, dev tools and self-compiling workflows

> **TL;DR** Three decision shapes, split from [[ideas/patterns-agents]] on 2026-09-25 with their IDs unchanged: P10 staged diff review and semantic lint, P11 test-output interpretation, P37 self-compiling workflows. Scan **Decision**, then **Map**. Agent loops, routing, gates, context and memory (P01-P09, P35, P36) stay on [[ideas/patterns-agents]].

## How to read this

One pattern = one narrow judgment. **Fit** is our verdict against [[concepts/jaggedness-jev-1-13]] and [[concepts/system-one]]: `strong` = bounded options, semantic judgment, code owns the rest (still starts with an experiment); `workable` = real, caveat named; `poor` = arithmetic, dates, indirection, generation or huge state. Pattern IDs are permanent; the full index is [[ideas/patterns]]. Posted numbers are attributed and unverified.

## Coding agents & dev tools

**P10 Staged diff review and semantic lint**
- *Decision* does this hunk carry a correctness, security or house-rule risk
- *State* one diff hunk, its context, and the rule text verbatim
- *Ask* `Noul` per risk class and per rule, plus severity `Score`, one request per hunk
- *Code* aggregate with `max`, not a mean; follow strong signals into a second pass
- *Fit* workable — literal reading means every rule needs its own precise question
- *Map* [[guides/writing-instructions-and-criteria]]
- *Seen* [@devagrawal09](https://x.com/devagrawal09/status/2100341005690298687), `devagrawal09/jev-review`, `jkudish/jev-mcp` (MCP tools: a four-rubric `Score` composited in code, plus an evidence gate on "tests pass" claims), [`valentynkit/jev-commit`](https://github.com/valentynkit/jev-commit) (five checks of message vs staged diff in one request; warns, and blocks only a credential on an added line; author: four cents per thousand commits), [`doeixd/jev-pref`](https://github.com/doeixd/jev-pref) (AGENTS.md preferences compiled into semantic lint rules), [`supercorp-ai/supercov`](https://github.com/supercorp-ai/supercov) (quality and security scores per file beside coverage, cached by content hash; vendor: about a cent per megabyte of source, so the source is read by TypeSafe's API (inferred)), [`0x7067/claude-jev`](https://github.com/0x7067/claude-jev) (rules read from the instruction files you already keep, one batched `Noul` per rule a hunk could break; blocks at ≥ 0.80, a second call with more context for 0.50-0.80; author, v0.21.0: 22 of 247 accepted real edits blocked, 17 of them by one repo's own no-comments rule, and 17/29 planted violations caught: a rule your team ignores will fire; [[ideas/builds-gates-and-routers]])

**P11 Test-output interpretation**
- *Decision* did tests pass, and is the failure related to this change
- *State* raw test output plus the diff
- *Ask* `Noul` (passed?) plus `Choice` over `unrelated_flake | related | environment`
- *Code* parse exit codes and counts in code; only attribution is a question
- *Fit* workable — counting is a documented weakness, so never ask "how many failed"
- *Map* [[concepts/jaggedness-jev-1-13]]
- *Seen* [@cyrilXBT](https://x.com/cyrilXBT/status/2101515130798297448) (~13s overhead down to ~0.83s)

## Self-compiling workflows

**P37 Self-compiling workflow: let the agent do the job, then fire itself from it**
- *Decision* per case, can the learned workflow handle this or does it go back to the full agent — plus every typed question inside it
- *State* the case record; inside the workflow, one item plus the criteria written from the procedure (Jev never sees the 508-line SOP)
- *Ask* four shapes — *Route* (`Choice` over branches plus an unsure branch), *Classify* (output schema → questions: enum → `Choice`, boolean → `Noul`), *Sift* (one `Noul` over every list item in one request), *Pick* (`Choice` of one item or none) — plus a `verify` clause checking an agent's submission field by field against the evidence it cited, and confidence predicates gating retries and escalations
- *Code* "Jev decides what a thing is; code decides what follows": ages, dates, tallies, two-strikes rules, and the share of cases routed to the workflow
- *Fit* strong — the decision/action seam, credited to 12-Factor Agents: an agent only where something must be found or done in the world, a typed question where something must be decided, code where mechanical. Compiling is a frontier agent's job, not Jev's
- *Map* [[patterns/intent-routing]], [[guides/testing-and-evaluation]], [[concepts/confidence]]
- *Seen* [@MiguelriosEN](https://x.com/MiguelriosEN/status/2101029313906987422) (Grep.ai's AgentRun article), announcement [@_aj](https://x.com/_aj/status/2102061534956662818)

*What to copy.* **Share growth**, scheduled in advance: of 1,000 AML alerts the first 100 ran the full agent for traces and notes; then 50% of the next 200 could try the workflow, 75%, 90%, all of the last 300 — ending 98 in 100, hand-backs ~1 in 20 early, ~1 in 70 late. **Validation**: a candidate ships only after beating the live version on a dev set, plus a reserved set the author never sees — he matched every dev label yet scored the same reserved. **Asymmetry survives compilation**: clearing a true match is the regulatory failure, so thresholds stay lopsided, not a rubric average. **Governance**: each decision names its node, model, inputs, questions and probabilities, answers kept — a verdict-rules rewrite replayed 868 hits in two minutes for a few cents, no model call. **Open problems**: the by-the-book agent took 826 tool calls and 51 minutes (two runs hit the hour unfinished); splitting too finely leaves a judge with half the policy; crash recovery skips model nodes and nested workflows.

*Their numbers, Grep.ai-reported and unverified:* ~$3/alert for a frontier agent ($2.89 production → $0.25 over 100 alerts); 50–95% savings; **100,000 alerts >$290K on Opus 5 versus <$26K with the harness**; their compliance judge scored 90% (first 100, full agent) to 95% (last 300). Jev claims **verified as consistent** with official pages, not independently measured: ~150 ms sits inside the 70–500 ms claim, and "a few hundred-thousandths of a dollar" per answer matches $0.042/Mtok input, output free (~700 tokens, [[reference/models-and-pricing]]).

## Related

- [[ideas/patterns]] — index, routing, anti-patterns; [[ideas/patterns-agents]] — agent internals, routing, gates, context, memory (P01-P09, P35, P36)
- [[ideas/builds-agents]], [[ideas/builds-gates-and-routers]] — builds behind these patterns; [[ideas/repos-agents]] — the repos in *Seen*
- [[ideas/consult]] — the "could Jev help this project?" conversation; [[ideas/measurements]] — the Grep.ai numbers beside other measurements

## Sources

Links are inline in each **Seen** line; the files behind them are in the frontmatter. P10, P11 and P37 moved here verbatim from ideas/patterns-agents on 2026-09-25; only the claude-jev entry in P10 is new.