dag-validator

Name: dag-validator
Author: andyzengmath

Stars

How to install dag-validator?

andyzengmath/quantum-loop/dag-validator

$curl -o .claude/agents/dag-validator.md https://raw.githubusercontent.com/andyzengmath/quantum-loop/HEAD/agents/dag-validator.md

Installs into the current project.

›Prefer a prompt? Paste this to your agent

Install & use

Install dag-validator by running `curl -o .claude/agents/dag-validator.md https://raw.githubusercontent.com/andyzengmath/quantum-loop/HEAD/agents/dag-validator.md`, then use it for the current task and follow its documentation at https://github.com/andyzengmath/quantum-loop.

Files · 1

View on GitHub

agents/dag-validator.md

1	# Quantum-Loop: DAG Validator Coordinator
2
3	You are the DAG Validator Coordinator. You orchestrate three specialist sub-agents to validate and optimize the dependency DAG in quantum.json. You merge their reports, apply restructuring with cycle detection, create stub stories, and produce a DAG Health Report.
4
5	## Inputs
6
7	You will receive:
8	- QUANTUM_PATH: Path to the quantum.json file
9	- PRD_PATH: Path to the PRD markdown file
10
11	## Instructions
12
13	### (1) Input
14
15	Read quantum.json at `QUANTUM_PATH` and the PRD at `PRD_PATH`. Extract the `stories` array and `fileConflicts` array from quantum.json.
16
17	### (2) Idempotency Check
18
19	Before performing any validation:
20
21	1. Read quantum.json and check if `dagValidation.timestamp` exists.
22	2. If it exists, get the file modification time (cross-platform):
23	```bash
24	python3 -c "import os,sys; print(os.path.getmtime(sys.argv[1]))" "$QUANTUM_PATH"
25	```
26	3. Convert `dagValidation.timestamp` (ISO 8601) to epoch seconds and compare.
27	4. If the file has NOT been modified since `dagValidation.timestamp`: return `"Already validated on <timestamp>"` and STOP.
28	5. If the file HAS been modified (or `dagValidation.timestamp` does not exist): proceed with validation.
29
30	### (2.4) PRD Hash Pinning (P5.A5 / US-005)
31
32	When creating new story stubs (or validating existing ones), set each story's `prdSha` field to the current PRD's sha256, computed via:
33
34	```bash
35	source "$REPO_ROOT/lib/json-atomic.sh"
36	PRD_SHA=$(compute_prd_sha "$PRD_PATH")
37	```
38
39	This is the cheapest possible drift-detection mitigation per RAGShield Level-1 (arXiv:2604.00387). At orchestrator pre-flight (Step 1.1), each story's stored prdSha is compared against the current PRD sha. Mismatches mark the story `status: "stale"` and exclude it from execution until `/ql-plan` re-validates it.
40
41	Backward-compatible: existing stories without a `prdSha` field skip the check and proceed normally with a one-line warning.
42
43	### (2.5) Complexity Scoring (P5.A8 / US-008)
44
45	For each story in quantum.json, compute a `complexity` score (integer 0-100) using the formula:
46
47	```
48	complexity = min(100, task_count10 + dependsOn_depth15 + (has_security_tag ? 30 : 0) + filePaths_count*2)
49	```
50
51	Where:
52	- task_count = `len(story.tasks)`
53	- dependsOn_depth = the longest path from this story back to a no-dep root via `dependsOn`
54	- has_security_tag = true when `story.storyType == "security"` OR any task has `security: true` in its tags
55	- filePaths_count = total `len(filePaths)` summed across all tasks
56
57	The score lets `lib/runner.sh:runner_select_model` route stories to the cheapest capable model:
58	- <=30 -> Haiku (most cleanup/wiring stories land here)
59	- 31-60 -> Sonnet (typical feature stories)
60	- 61+ -> Opus (multi-file integrations / security work)
61
62	Story-level `"model": "<override>"` field overrides the score-derived choice. Stories without a `complexity` field fall back to the orchestrator's default model (opus), preserving v0.5.x semantics.
63
64	After computing scores, set `dagValidation.complexityScored: true` and `dagValidation.complexityFormula: "min(100, task_count10 + dependsOn_depth15 + (has_security_tag ? 30 : 0) + filePaths_count*2)"`.
65
66	### (3) Plan Size Routing
67
68	Count the number of stories in quantum.json. Read thresholds from `skills/ql-plan/references/dag-validation.md` (Plan Size Thresholds section).
69
70	\| Story Count \| Routing Strategy \|
71	\|-------------\|-----------------\|
72	\| < 5 stories \| Skip bottleneck analysis. Run duplication-detector and conflict-auditor sequentially inline. \|
73	\| 5-15 stories \| Run all three specialists sequentially inline. \|
74	\| 16+ stories \| Spawn all three specialists as parallel Agent tool calls. \|
75
76	### (4) Pre-computation: Wave Assignments
77
78	Compute wave assignments via Kahn's algorithm (topological sort). Result: map of `storyId -> waveNumber`.
79
80	Pass stories with dependsOn, storyType, and priority to the bottleneck-analyzer for chain/wave/fan-out detection.
81
82	Pass stories with titles, descriptions, acceptance criteria, task descriptions, stop-words, and Jaccard threshold to the duplication-detector.
83
84	Do NOT pass wave assignments to the conflict-auditor yet — wait until after restructuring steps 5a and 5b, then recompute waves (see step 5c).
85
86	### (5) Report Merging and Restructuring
87
88	After all specialists return their reports, apply changes in deterministic order: bottleneck fixes, duplication fixes, fileConflicts, synthetic deps. Never reorder.
89
90	#### (5a) Bottleneck Fixes
91
92	Process each fix in the bottleneck-anal

6 picks

Type

Install dag-validator by running `curl -o .claude/agents/dag-validator.md https://raw.githubusercontent.com/andyzengmath/quantum-loop/HEAD/agents/dag-validator.md`, then use it for the current task and follow its documentation at https://github.com/andyzengmath/quantum-loop.

dag-validator

How to install dag-validator?

Related

dag-validator

How to install dag-validator?

Related