Comparison

A · Cline rules · kurnoolion/nora B · CLAUDE.md · kurnoolion/nora

What each file covers, counted
Dimension	Only in A	Only in B	Overlap
Sections	5	8	0%
Commands	1	1	0%
Section tags	1	4	0%

What each file covers

Sections

0 shared · 5 only in A · 8 only in B

− Your role: on-prem student to a Teacher LLM
− The standard loop
− What you do
− What you do NOT do
− Per-session
+ CLAUDE.md
+ Project
+ Session-start ritual
+ Key skills
+ State files
+ Current state
+ Repository layout
+ Conventions

Commands

0 shared · 1 only in A · 1 only in B

− git pull
+ python -m core.src.<module>.<module>_cli

Section tags

0 shared · 1 only in A · 4 only in B

− do-not
+ test
+ code-style
+ architecture
+ agent-behaviour

Line diff

+53 added−41 removed11 unchanged17.2% identical

kurnoolion/nora · .clinerules/01-role.md

@@ −1 @@

1# Your role: on-prem student to a Teacher LLM

3The user works with two AI partners:

4- **Teacher LLM** — sees the full repo, designs and codes; cannot see internal corpus.

5- **You (Cline)** on the on-prem PC — sees the internal corpus under `<env_dir>/input/`;

6 does NOT design or write code under `core/src/`.

8## The standard loop

10```

11 ┌──── on-prem (you + corpus) ──┐ ┌──── Teacher LLM ────────┐

12 │ │ manual │ │

13 │ 1. user invokes a playbook │ typing │ 3. read report │

14 │ 2. you produce a compact │ ───────────▶ │ 4. design + code │

15 │ redacted report │ │ 5. commit to git │

16 │ 6. user runs `git pull` │ ◀──── git ──── │ │

17 │ 7. you run new code │ │ │

18 │ 8. you produce next report │ ───────────▶ │ 9. respond │

19 └──────────────────────────────┘ └────────────────────────┘

20```

22Steps 3 + 9 are the user reading your screen and **hand-typing** the redacted version into

23Teacher LLM. Code never moves through chat — it moves through git.

25## What you do

27- Read the corpus (under `<env_dir>/input/`), profile docs, derive detection rules, run

28 the pipeline, capture stats.

29- Write to:

30 - `<env_dir>/state/cline-mapping.json` — your redaction mapping (on-prem only, never in git)

31 - `<env_dir>/reports/` — full reports (kept on-prem; user reads off your screen)

32 - `customizations/profiles/<plan>/profile.json` — per-document parser profiles (in repo)

33 - `customizations/corrections/` and `<env_dir>/corrections/` — correction files

34- Run NORA CLIs (`profile_debug`, `parser_cli`, `parse_review`, `pipeline.run_cli`,

35 `vectorstore_cli`, `query_cli`, `retrieval_debug`, `llm_debug`, `embed_debug`).

36- Apply Teacher LLM's commits via `git pull`.

38## What you do NOT do

40- Write Python code under `core/src/` — that's Teacher LLM's job, delivered via git.

41- Generate prompts, templates, or text content based on what's in the corpus.

42- Create reports longer than ~30 lines (the user has to hand-type them; longer ⇒ unusable).

43- Send any verbatim corpus content out — see `02-content-safety.md`.

44- Commit to `customizations/` if the change is mechanical and Teacher LLM should produce it

45 (e.g., schema changes); commit `customizations/profiles/<plan>/` if the change is

46 corpus-derived (regex tightening, applicability lists, definitions overrides).

48## Per-session

50On first conversation each session, run `cline-playbooks/orient.md` to load project context.

51Then proceed to the task at hand.

kurnoolion/nora · CLAUDE.md

@@ +1 @@

1# CLAUDE.md

3This file provides guidance to Claude Code (and Cline) when working with code in this repository.

5## Project

7**NORA** — Network Operator Requirements Analyzer. AI system combining a unified Knowledge Graph with targeted RAG for intelligent querying, cross-referencing, and compliance analysis of US MNO device requirement specifications across multiple MNOs (MNO-A, AT&T, T-Mobile) and quarterly releases.

9See `docs/compact/PROJECT.md` for the 1-page identity and `TDD_Telecom_Requirements_AI_System.md` for the canonical technical design.

11## Session-start ritual

13This project uses **COMPACT** — a portable scaffold for team AI-partnered software development. The AI scaffolding lives under `.claude/skills/` and the durable state lives under `docs/compact/`.

15At the start of every session, invoke `run the session-start skill` (or `/session-start`). It loads Tier 1 context (`PROJECT.md` / `STATUS.md` / `MAP.md` / the active phase prompt) and asks what you're working on. Re-invoke mid-session after Claude Code auto-compaction or any time context feels stale.

17## Key skills

19- `/session-start` — hydrate context at start of every session

20- `/switch-phase <requirements | architecture | development>` — adopt the phase persona

21- `/regen-map` — regenerate `MAP.md` + Structure sections when code structure changes

22- `/drift-check <requirements | design | dev-full | dev-module <name> | all>` — audit for drift between requirements, design, and implementation

23- `/close-session` — end-of-session ritual: triages decisions, updates STATUS, audits MODULE.md edits, proposes commit (this is the **only** place memory is made)

24- `/project-init --re-init` — regenerate phase prompts after project-level changes (state files preserved)

26## State files

28- `docs/compact/PROJECT.md` — identity + Contributors table

29- `docs/compact/STATUS.md` — active phase, done / in progress / next, flags

30- `docs/compact/requirements.md` — FR / NFR / Deferred (authority for what the system must do)

31- `docs/compact/MAP.md` — module table + Mermaid dependency graph (regenerated)

32- `docs/compact/DECISIONS.md` — append-only ADR log

33- `docs/compact/structure-conventions.md` — what's a module, visibility mapping

34- `docs/compact/design-inputs/` — TDD, README, SESSION_SUMMARY, SETUP_OFFLINE (archival design inputs)

35- `docs/compact/retrofit-snapshot.md` — archival scan of existing codebase at retrofit time

36- `docs/compact/phases/{requirements,architecture,development}.md` — phase personas loaded by `/switch-phase`

37- `src/<module>/MODULE.md` — per-module contracts (16 skeletons seeded by retrofit — see STATUS for curation progress)

39## Current state

41Retrofitted on 2026-04-21 via `/project-init --retrofit`. The 16 MODULE.md skeletons were curated through 2026-04-27 (sentinels removed). Three-tier code reorg (`core/` + `customizations/` + `config/`) executed 2026-04-27. See `docs/compact/STATUS.md` for the current active phase and work list.

43## Repository layout

45Three-tier code organization (D-019..D-024) plus a per-env runtime directory `<env_dir>`. See `docs/compact/structure-conventions.md` for the canonical layout.

47- `core/src/` — AI-generated Python source (16 packages; one MODULE.md per package). Manual edits exceptional (D-019).

48- `core/tests/` — pytest suite (one `test_<module>.py` per package).

49- `customizations/profiles/` — human-curated document profiles (committed JSON).

50- `customizations/llm/` — AI-scaffolded LLM provider boilerplate (e.g. `proprietary_provider.py`); humans complete the `complete()` body for production deployments.

51- `config/` — per-module settings (currently `config/web.json`; more files land as modules surface configurable knobs).

52- `environments/` — per-environment configs (gitignored except `.gitkeep`); each names an `env_dir` path.

53- `docs/compact/` — COMPACT state files.

54- `.claude/skills/` — COMPACT skills (flat layout: compact-skill, session-start, switch-phase, regen-map, drift-check, close-session, project-init, doctor).

55- `<env_dir>/` (runtime, user-supplied path; not in repo) — `input/<MNO>/<release>/`, `out/<stage>/`, `state/`, `corrections/`, `reports/`, `eval/`.

57## Conventions

59- Python; no `pyproject.toml` — `requirements.txt` + `core/src/` layout with `__init__.py` per package.

60- Public surface: top-level identifiers without a leading underscore (plus `__init__.py` re-exports when `__all__` is used).

61- CLI per module: `core/src/<module>/<module>_cli.py` with `main()` entrypoint; invoked as `python -m core.src.<module>.<module>_cli`.

62- Protocol-based abstractions: `LLMProvider`, `EmbeddingProvider`, `VectorStoreProvider`.

63- No proprietary document content in logs, error messages, compact reports, or test fixtures.

@@ −1 +1 @@

1−# Your role: on-prem student to a Teacher LLM

1+# CLAUDE.md

3−The user works with two AI partners:

4−- **Teacher LLM** — sees the full repo, designs and codes; cannot see internal corpus.

5−- **You (Cline)** on the on-prem PC — sees the internal corpus under `<env_dir>/input/`;

6− does NOT design or write code under `core/src/`.

3+This file provides guidance to Claude Code (and Cline) when working with code in this repository.

8−## The standard loop

5+## Project

10−```

11− ┌──── on-prem (you + corpus) ──┐ ┌──── Teacher LLM ────────┐

12− │ │ manual │ │

13− │ 1. user invokes a playbook │ typing │ 3. read report │

14− │ 2. you produce a compact │ ───────────▶ │ 4. design + code │

15− │ redacted report │ │ 5. commit to git │

16− │ 6. user runs `git pull` │ ◀──── git ──── │ │

17− │ 7. you run new code │ │ │

18− │ 8. you produce next report │ ───────────▶ │ 9. respond │

19− └──────────────────────────────┘ └────────────────────────┘

20−```

7+**NORA** — Network Operator Requirements Analyzer. AI system combining a unified Knowledge Graph with targeted RAG for intelligent querying, cross-referencing, and compliance analysis of US MNO device requirement specifications across multiple MNOs (MNO-A, AT&T, T-Mobile) and quarterly releases.

218

22−Steps 3 + 9 are the user reading your screen and **hand-typing** the redacted version into

23−Teacher LLM. Code never moves through chat — it moves through git.

9+See `docs/compact/PROJECT.md` for the 1-page identity and `TDD_Telecom_Requirements_AI_System.md` for the canonical technical design.

2410

25−## What you do

11+## Session-start ritual

2612

27−- Read the corpus (under `<env_dir>/input/`), profile docs, derive detection rules, run

28− the pipeline, capture stats.

29−- Write to:

30− - `<env_dir>/state/cline-mapping.json` — your redaction mapping (on-prem only, never in git)

31− - `<env_dir>/reports/` — full reports (kept on-prem; user reads off your screen)

32− - `customizations/profiles/<plan>/profile.json` — per-document parser profiles (in repo)

33− - `customizations/corrections/` and `<env_dir>/corrections/` — correction files

34−- Run NORA CLIs (`profile_debug`, `parser_cli`, `parse_review`, `pipeline.run_cli`,

35− `vectorstore_cli`, `query_cli`, `retrieval_debug`, `llm_debug`, `embed_debug`).

36−- Apply Teacher LLM's commits via `git pull`.

13+This project uses **COMPACT** — a portable scaffold for team AI-partnered software development. The AI scaffolding lives under `.claude/skills/` and the durable state lives under `docs/compact/`.

3714

38−## What you do NOT do

15+At the start of every session, invoke `run the session-start skill` (or `/session-start`). It loads Tier 1 context (`PROJECT.md` / `STATUS.md` / `MAP.md` / the active phase prompt) and asks what you're working on. Re-invoke mid-session after Claude Code auto-compaction or any time context feels stale.

3916

40−- Write Python code under `core/src/` — that's Teacher LLM's job, delivered via git.

41−- Generate prompts, templates, or text content based on what's in the corpus.

42−- Create reports longer than ~30 lines (the user has to hand-type them; longer ⇒ unusable).

43−- Send any verbatim corpus content out — see `02-content-safety.md`.

44−- Commit to `customizations/` if the change is mechanical and Teacher LLM should produce it

45− (e.g., schema changes); commit `customizations/profiles/<plan>/` if the change is

46− corpus-derived (regex tightening, applicability lists, definitions overrides).

17+## Key skills

4718

48−## Per-session

19+- `/session-start` — hydrate context at start of every session

20+- `/switch-phase <requirements | architecture | development>` — adopt the phase persona

21+- `/regen-map` — regenerate `MAP.md` + Structure sections when code structure changes

22+- `/drift-check <requirements | design | dev-full | dev-module <name> | all>` — audit for drift between requirements, design, and implementation

23+- `/close-session` — end-of-session ritual: triages decisions, updates STATUS, audits MODULE.md edits, proposes commit (this is the **only** place memory is made)

24+- `/project-init --re-init` — regenerate phase prompts after project-level changes (state files preserved)

4925

50−On first conversation each session, run `cline-playbooks/orient.md` to load project context.

51−Then proceed to the task at hand.

26+## State files

27+

28+- `docs/compact/PROJECT.md` — identity + Contributors table

29+- `docs/compact/STATUS.md` — active phase, done / in progress / next, flags

30+- `docs/compact/requirements.md` — FR / NFR / Deferred (authority for what the system must do)

31+- `docs/compact/MAP.md` — module table + Mermaid dependency graph (regenerated)

32+- `docs/compact/DECISIONS.md` — append-only ADR log

33+- `docs/compact/structure-conventions.md` — what's a module, visibility mapping

34+- `docs/compact/design-inputs/` — TDD, README, SESSION_SUMMARY, SETUP_OFFLINE (archival design inputs)

35+- `docs/compact/retrofit-snapshot.md` — archival scan of existing codebase at retrofit time

36+- `docs/compact/phases/{requirements,architecture,development}.md` — phase personas loaded by `/switch-phase`

37+- `src/<module>/MODULE.md` — per-module contracts (16 skeletons seeded by retrofit — see STATUS for curation progress)

38+

39+## Current state

40+

41+Retrofitted on 2026-04-21 via `/project-init --retrofit`. The 16 MODULE.md skeletons were curated through 2026-04-27 (sentinels removed). Three-tier code reorg (`core/` + `customizations/` + `config/`) executed 2026-04-27. See `docs/compact/STATUS.md` for the current active phase and work list.

42+

43+## Repository layout

44+

45+Three-tier code organization (D-019..D-024) plus a per-env runtime directory `<env_dir>`. See `docs/compact/structure-conventions.md` for the canonical layout.

46+

47+- `core/src/` — AI-generated Python source (16 packages; one MODULE.md per package). Manual edits exceptional (D-019).

48+- `core/tests/` — pytest suite (one `test_<module>.py` per package).

49+- `customizations/profiles/` — human-curated document profiles (committed JSON).

50+- `customizations/llm/` — AI-scaffolded LLM provider boilerplate (e.g. `proprietary_provider.py`); humans complete the `complete()` body for production deployments.

51+- `config/` — per-module settings (currently `config/web.json`; more files land as modules surface configurable knobs).

52+- `environments/` — per-environment configs (gitignored except `.gitkeep`); each names an `env_dir` path.

53+- `docs/compact/` — COMPACT state files.

54+- `.claude/skills/` — COMPACT skills (flat layout: compact-skill, session-start, switch-phase, regen-map, drift-check, close-session, project-init, doctor).

55+- `<env_dir>/` (runtime, user-supplied path; not in repo) — `input/<MNO>/<release>/`, `out/<stage>/`, `state/`, `corrections/`, `reports/`, `eval/`.

56+

57+## Conventions

58+

59+- Python; no `pyproject.toml` — `requirements.txt` + `core/src/` layout with `__init__.py` per package.

60+- Public surface: top-level identifiers without a leading underscore (plus `__init__.py` re-exports when `__all__` is used).

61+- CLI per module: `core/src/<module>/<module>_cli.py` with `main()` entrypoint; invoked as `python -m core.src.<module>.<module>_cli`.

62+- Protocol-based abstractions: `LLMProvider`, `EmbeddingProvider`, `VectorStoreProvider`.

63+- No proprietary document content in logs, error messages, compact reports, or test fixtures.

5264