wip: preserve local engine and agent workflow changes before ganjihong refactor integration

backend/README.md

@@ -20,53 +20,36 @@ CDSL_OPENAI_REASONING_EFFORT=medium
 ```
 
 Use `low`, `medium`, or `high` according to the latency/cost versus quality
-tradeoff. The setting is sent as Chat Completions' `reasoning_effort` field to
-authoring, streaming, and visual-review requests. Leave it empty to use the
+tradeoff. The setting is sent as the provider's `reasoning_effort` field to
+requirements analysis and Authoring CDSL generation. Leave it empty to use the
 provider/model default. The selected OpenAI-compatible endpoint must support
 the requested value.
 
 ## Autonomous CDSL Agent Configuration
 
-The autonomous agent writes one frozen free-form `requirements.md`, then
-observes, measures, renders and appends one CDSL feature at a time. Its author
-uses normal function calls; no provider strict JSON Schema capability or
-complete modelling DAG is required. Candidate fragments are rebuilt in a
-staging directory through `cdsl_only` before a checkpoint can be committed.
+Each task has one bounded workflow:
 
-Final publication requires a separately configured vision-capable review model
-and the Python OpenCascade/Pillow technical renderer. The agent may build and
-inspect intermediate checkpoints without image review; a final run fails
-closed if its independent review configuration is unavailable.
+```text
+request analysis -> complete cad.author.v1 -> server compilation -> runtime build
+-> at most two complete repairs -> final or best-effort publication
+```
+
+The model outputs only local body/feature names and declarative selectors.
+The server validates strict schemas, allocates Runtime CDSL identities,
+compiles references, executes dependencies, and preserves the last executable
+prefix. STEP is the primary artifact; GLB and the CPU-only OpenCascade/Pillow
+render bundle are generated from the same published revision.
 
 ```dotenv
-# Must name one configured provider and one model listed in that provider's
-# CDSL_<PROVIDER>_VISION_MODELS setting. It is intentionally not inferred
-# from the authoring model.
-CDSL_REVIEW_PROVIDER=deepseek
-CDSL_REVIEW_MODEL=deepseek-v4-flash-vision-exp
-CDSL_DEEPSEEK_VISION_MODELS=deepseek-v4-flash-vision-exp
-
 # Install Python rendering dependencies. The renderer reads the revision STEP
 # file and creates canonical images without a browser or GPU driver.
 pip install -r requirements.txt
-
-# Limits apply to the current checkpoint head, never to total task complexity.
-CDSL_AGENT_TOOL_CALLS_PER_CYCLE=12
-CDSL_AGENT_CANDIDATE_ATTEMPTS_PER_HEAD=3
-CDSL_AGENT_CONSECUTIVE_NO_PROGRESS_LIMIT=6
-CDSL_AGENT_MAX_FEATURES_PER_FRAGMENT=6
-CDSL_AGENT_CONTEXT_CHAR_LIMIT=24000
-CDSL_AGENT_RENDER_CACHE=true
 ```
 
-The author chooses each coherent 1-6 feature batch. Every rebuilt batch is
-rendered and independently reviewed before it can become a checkpoint; only
-an accepted reviewer verdict advances the working model. Every checkpoint is rebuilt from its fully materialized CDSL through the
-`cdsl_only` runtime. Checkpoint GLB files are preview-only; STEP, CDSL, and
-reports are available only after the task reaches `COMPLETED`.
-
-The backend assigns feature and sketch IDs, appends causal dependencies and
-expands only opaque current-snapshot selector tokens. It does not compile
-geometry templates or correct workplanes, profiles, sizes, directions or
-boolean semantics authored by the model. Failed candidates remain auditable
-but never become revisions.
+The initial generation plus two repairs are the only model calls allowed after
+requirements analysis. A repair returns a complete replacement Authoring CDSL
+and may only alter diagnosed features. Runtime selector ambiguity, missing
+selectors, and unavailable dependencies produce stable diagnostics rather than
+topology guesses. Requirement compliance is reported independently as
+`pass`, `fail`, `pending`, or `not_applicable`; a partial executable model is
+still published after the repair budget is exhausted.