Forward implementation first
Build working capability and correct output before administrative bookkeeping.
Apply this contract both while building a system and while running it.
Scope
This skill applies to:
- infrastructure and roadmap implementation;
- input intake and intermediate artifact production;
- transformation, derivation, and analysis stages;
- test construction, runtime verification, and measurement;
- publication of final output and closeout;
- manual invocation of any authorized pipeline stage.
It applies whether the action runs through a full orchestrator, a focused
command, or a manual stage invocation.
Decision rule
Before each action, classify it as one of:
- Semantic implementation: builds or connects a producer, consumer,
adapter, runtime path, schema, fixture, or final output.
- Focused validation: tests the changed dependency cone through behavior,
schema, counts, samples, conservation, consistency, nontruncation, or
measured resources.
- Administrative bookkeeping: generates or repairs hashes, locks,
receipts, dashboards, certification markers, progress metadata, or
presence-only records.
Choose categories 1 and 2. Skip category 3 unless the user asks for it or the
artifact is itself part of the product. When administrative work blocks a path
without protecting correctness, remove that dependency from the path.
Hard constraints
- Build the producer and consumer before polishing status or certification
surfaces.
- Do not generate, repair, compare, or propagate administrative hashes.
- Do not create or wait on filesystem locks.
- Do not rerun an unchanged stage to regenerate a receipt or marker.
- Never invalidate valid output because an administrative receipt, hash,
certification marker, dashboard row, or progress record is missing, stale, or
incompatible.
- Never move the forward cursor backward for an administrative metadata change.
- Never invalidate a wide stage range when only one producer or consumer
changed. Replay only that producer's dependency cone.
- Do not treat certification, receipts, dashboards, progress metadata, or
artifact presence as the product.
- Remove hash-only, lock-only, receipt-only, and presence-only credit or gates
from execution paths.
- Do not claim a capability works because its file exists. Run the smallest
changed dependency cone and inspect the output.
- Publish valid output after focused validation. Do not add an extra review
cycle when no defect remains.
Forward cursor and manual execution
A stage may move backward or be replayed only when at least one real condition
holds:
- its input meaning changed;
- its target or pinned revision changed;
- its output is malformed, truncated, nonconserving, internally inconsistent,
or incompatible with the consumer;
- an observed run disproves the earlier static result;
- the changed producer's declared dependency cone requires replay.
Missing or stale administrative metadata is not one of those conditions.
When the orchestrator refuses to run a stage only because of a receipt,
certification marker, progress record, administrative hash, or lock:
- Run the exact stage manually.
- Validate its output with the focused checks below.
- Publish the valid output atomically.
- Continue from the forward cursor.
- Remove or downgrade the administrative-only gate in the execution path.
Do not refuse an authorized manual run because the full pipeline cannot issue a
receipt. Do not push the user toward another agent or tool to get around your
own bookkeeping.
Required validation
Use the checks that match the change:
- runtime behavior and exit status;
- schema and type validity;
- exact input, output, accepted, rejected, and unknown counts;
- deterministic first, middle, and last samples;
- identity and partition conservation;
- join consistency and flag polarity;
- nontruncation and bounded diagnostic output;
- wall-clock time and peak memory for material stages.
Hashes may identify inputs or revisions, but they never grant correctness,
execution, or roadmap credit.
Exceptions that stay substantive
Bookkeeping is cheap to skip. Evidence is not. Keep these:
- Integrity that belongs to the product itself. A checksum your users verify, a
signature your format requires, a hash that is part of the output contract.
Those are features, not paperwork.
- Input and revision identity when it decides which version of the thing you
are operating on. Getting that wrong means correct work on the wrong target.
- Real measurement: tests that assert results, benchmarks, reproductions,
end-to-end runs, and whatever your domain treats as proof.
- The difference between coverage and consequence. A run that exercises a path
is not a check that the path produced the right answer.
- An execution record is substantive when it carries the command, the input,
the result, and the expectation it was checked against. Its absence blocks
that specific claim. It does not invalidate unrelated earlier stages.
Execution discipline
- Run one heavy process at a time: compiler, full test suite, large data job,
benchmark, long scan, or anything else that saturates CPU, memory, or a port.
- Keep parallel workers on light, nonoverlapping implementation and
focused-test lanes.
- When the operator authorizes a worker pool of size N and useful bounded
backlog exists, keep up to N cheap-model lanes assigned continuously. Refill
a completed lane with the next independent light task without waiting for the
whole wave. Do not manufacture busywork to occupy slots.
- The root lane alone owns heavy execution, publication, cursor movement,
acceptance, and conclusions. Parallel lanes prepare, implement, inspect, and
test nonoverlapping support work. They never become competing truth.
- Forward progress must not wait for every parallel lane to finish. Consume
completed lane output after local trust-but-verify checks, when its owning
dependency is reached.
- Replay the smallest affected dependency cone from the forward cursor.
- If a heavy run exposes a defect in a stage, fix that stage before rerunning
the cone.
- If the next proposed action is bookkeeping-only, select the next unresolved
producer, consumer, adapter, or output instead.
Status reporting
Report implemented behavior and measured output first. List blockers literally.
Keep infrastructure progress separate from evidence about the output. Do not
turn administrative completion into a substitute for working capability.
1---2name: forward-implementation-first3description: >4 Keeps an agent building and validating real output instead of servicing its5 own bookkeeping. Use for multi-stage pipelines, capability roadmaps,6 long-running implementation loops, migrations, staged data or build runs, and7 any run where administrative hashes, locks, receipts, dashboards,8 certification markers, or progress metadata can block correct work. Use when9 an agent refuses to advance, rewinds finished stages, or reruns unchanged10 work because a marker is missing or stale.11metadata:12 source: https://skillmd.com/skills/openclaw-cn-dev/forward-implementation-first13---1415# Forward implementation first1617Build working capability and correct output before administrative bookkeeping.18Apply this contract both while building a system and while running it.1920## Scope2122This skill applies to:2324- infrastructure and roadmap implementation;25- input intake and intermediate artifact production;26- transformation, derivation, and analysis stages;27- test construction, runtime verification, and measurement;28- publication of final output and closeout;29- manual invocation of any authorized pipeline stage.3031It applies whether the action runs through a full orchestrator, a focused32command, or a manual stage invocation.3334## Decision rule3536Before each action, classify it as one of:37381. **Semantic implementation**: builds or connects a producer, consumer,39 adapter, runtime path, schema, fixture, or final output.402. **Focused validation**: tests the changed dependency cone through behavior,41 schema, counts, samples, conservation, consistency, nontruncation, or42 measured resources.433. **Administrative bookkeeping**: generates or repairs hashes, locks,44 receipts, dashboards, certification markers, progress metadata, or45 presence-only records.4647Choose categories 1 and 2. Skip category 3 unless the user asks for it or the48artifact is itself part of the product. When administrative work blocks a path49without protecting correctness, remove that dependency from the path.5051## Hard constraints5253- Build the producer and consumer before polishing status or certification54 surfaces.55- Do not generate, repair, compare, or propagate administrative hashes.56- Do not create or wait on filesystem locks.57- Do not rerun an unchanged stage to regenerate a receipt or marker.58- Never invalidate valid output because an administrative receipt, hash,59 certification marker, dashboard row, or progress record is missing, stale, or60 incompatible.61- Never move the forward cursor backward for an administrative metadata change.62- Never invalidate a wide stage range when only one producer or consumer63 changed. Replay only that producer's dependency cone.64- Do not treat certification, receipts, dashboards, progress metadata, or65 artifact presence as the product.66- Remove hash-only, lock-only, receipt-only, and presence-only credit or gates67 from execution paths.68- Do not claim a capability works because its file exists. Run the smallest69 changed dependency cone and inspect the output.70- Publish valid output after focused validation. Do not add an extra review71 cycle when no defect remains.7273## Forward cursor and manual execution7475A stage may move backward or be replayed only when at least one real condition76holds:7778- its input meaning changed;79- its target or pinned revision changed;80- its output is malformed, truncated, nonconserving, internally inconsistent,81 or incompatible with the consumer;82- an observed run disproves the earlier static result;83- the changed producer's declared dependency cone requires replay.8485Missing or stale administrative metadata is not one of those conditions.8687When the orchestrator refuses to run a stage only because of a receipt,88certification marker, progress record, administrative hash, or lock:89901. Run the exact stage manually.912. Validate its output with the focused checks below.923. Publish the valid output atomically.934. Continue from the forward cursor.945. Remove or downgrade the administrative-only gate in the execution path.9596Do not refuse an authorized manual run because the full pipeline cannot issue a97receipt. Do not push the user toward another agent or tool to get around your98own bookkeeping.99100## Required validation101102Use the checks that match the change:103104- runtime behavior and exit status;105- schema and type validity;106- exact input, output, accepted, rejected, and unknown counts;107- deterministic first, middle, and last samples;108- identity and partition conservation;109- join consistency and flag polarity;110- nontruncation and bounded diagnostic output;111- wall-clock time and peak memory for material stages.112113Hashes may identify inputs or revisions, but they never grant correctness,114execution, or roadmap credit.115116## Exceptions that stay substantive117118Bookkeeping is cheap to skip. Evidence is not. Keep these:119120- Integrity that belongs to the product itself. A checksum your users verify, a121 signature your format requires, a hash that is part of the output contract.122 Those are features, not paperwork.123- Input and revision identity when it decides which version of the thing you124 are operating on. Getting that wrong means correct work on the wrong target.125- Real measurement: tests that assert results, benchmarks, reproductions,126 end-to-end runs, and whatever your domain treats as proof.127- The difference between coverage and consequence. A run that exercises a path128 is not a check that the path produced the right answer.129- An execution record is substantive when it carries the command, the input,130 the result, and the expectation it was checked against. Its absence blocks131 that specific claim. It does not invalidate unrelated earlier stages.132133## Execution discipline134135- Run one heavy process at a time: compiler, full test suite, large data job,136 benchmark, long scan, or anything else that saturates CPU, memory, or a port.137- Keep parallel workers on light, nonoverlapping implementation and138 focused-test lanes.139- When the operator authorizes a worker pool of size N and useful bounded140 backlog exists, keep up to N cheap-model lanes assigned continuously. Refill141 a completed lane with the next independent light task without waiting for the142 whole wave. Do not manufacture busywork to occupy slots.143- The root lane alone owns heavy execution, publication, cursor movement,144 acceptance, and conclusions. Parallel lanes prepare, implement, inspect, and145 test nonoverlapping support work. They never become competing truth.146- Forward progress must not wait for every parallel lane to finish. Consume147 completed lane output after local trust-but-verify checks, when its owning148 dependency is reached.149- Replay the smallest affected dependency cone from the forward cursor.150- If a heavy run exposes a defect in a stage, fix that stage before rerunning151 the cone.152- If the next proposed action is bookkeeping-only, select the next unresolved153 producer, consumer, adapter, or output instead.154155## Status reporting156157Report implemented behavior and measured output first. List blockers literally.158Keep infrastructure progress separate from evidence about the output. Do not159turn administrative completion into a substitute for working capability.