ECSS Harness Manufacturing — Complementary Labelling and Marking (space-systems/ecss/q2030-comp-labelling)
Use when the task is the complementary labelling layer of ECSS-Q-ST-20-30C
clause 7.6 -- the space-specific demands on what a harness marking says,
how it is formed and how long it stays readable, on top of the workmanship
marking criteria the standard delegates to its adopted chapter.
Domain quick reference
- A harness label is a traceability token, not a description. Its value
comes from decomposing cleanly into the fields the project format
prescribes, so a label that cannot be parsed has already failed however
sensible it reads to a person standing at the bench.
- Content is graded field by field, because each position carries its own
character set and length. A project code in lower case, a harness number
without its prefix letter or a serial carrying a letter all parse and
all still fail, and naming which field failed is what makes the finding
actionable.
- Character height is not an absolute. It scales with the distance the
marking has to be read from, so the same marking that is comfortable on
a bench harness is illegible on an installed one behind a bracket, and
the read distance belongs in the requirement rather than in the
inspector's judgement.
- Placement is a window, not a preference. Too close to the termination
and the marking is under the backshell or the strain relief; too far
down the wire and it is lost inside the bundle at the first breakout.
- A wire that disappears into a bundle owes marking at both ends. One end
identifies the wire at the connector it is being worked on; the far end
is what makes a fault traceable without unlacing the harness.
- Durability is evidence, not a material claim. The marking survived a
recorded number of abrasion and solvent exposures while staying legible,
and those counts are what is graded against the project requirement.
- Uniqueness is the whole point. Two wires carrying the same token are
worse than two unmarked wires, because the marking now actively
misdirects the next person to trace them.
Workflow
- Validate each label record: an identifier, the token, the character
height, the read distance and the distance from its termination.
- Decompose the token against the prescribed field order and refuse a
token whose field count does not match; an empty field is a refusal
too, not an optional position.
- Grade every parsed field against the character set and length its
position carries, naming the field in the finding.
- Compute the minimum character height from the read distance and grade
the actual height against it, absorbing an exact equality with a
tolerance instead of rounding the height up.
- Grade the distance from the termination against the placement window
at both edges.
- Where the wire is routed through a bundle, confirm both ends carry the
marking.
- Grade the abrasion and solvent legibility counts against the project
requirement and report each shortfall with its margin.
- Collect every token in the harness, report any carried more than once,
and roll all findings into one verdict.
Pitfalls
- Accepting a label because a person could read it. The format is what
makes the token machine-traceable through the as-built record; human
legibility is necessary and nowhere near sufficient.
- Grading the token as one string. A single regular expression over the
whole label tells the operator it is wrong without telling them which
field to fix, and the rework comes back wrong twice.
- Using one minimum character height everywhere. The requirement follows
the read distance, so an installed harness read from half a metre away
needs taller marking than the same wire on the bench.
- Treating placement as tidiness. A label under the backshell cannot be
read without demating, and a label past the first breakout is on the
wrong side of it.
- Marking only the connector end of a bundled wire. The far end is the
one a fault investigation needs, and adding it later means unlacing.
- Claiming durability from the marker material's datasheet. The
requirement is graded on the legibility that survived recorded
exposure, and a substitution of marker or solvent invalidates it.
- Letting a duplicate token through because both wires are in different
connectors. The token is unique across the harness; scoping uniqueness
to a connector is how two wires end up indistinguishable in the as-built
record.
Behavior contract (gate 3)
The token decomposition, the per-field content grading, the read-distance
character-height scaling, the placement-window grading, the both-end
marking rule, the abrasion and solvent durability verdict, the duplicate
token search and the whole-set roll-up are exercised by the gate 3
contract test: scripts/test_q2030_comp_labelling.py against
scripts/q2030_comp_labelling_logic.py (stdlib unittest, offline). Run:
python3 scripts/test_q2030_comp_labelling.py
Compliance
- ECSS standards are freely downloadable (ESA); cite the source and
paraphrase per standards-map.yaml.
- compliance: STANDARDS-REF, gated: false.
1---2name: q2030-comp-labelling3description: "Audit harness labelling and marking against the complementary content, format and durability requirements of ECSS-Q-ST-20-30C clause 7.6. Use when the task is decomposing a label token into the fields the project format prescribes, grading each field against the character set its position carries, scaling the minimum character height to the distance the marking is read from, placing the label inside its window from the termination, requiring both ends of a wire routed through a bundle to be marked, grading abrasion and solvent legibility evidence, and catching a token carried by two wires. Trigger: ecss, q-st-20-30c, complementary-harness-labelling, harness-label-token-format, marking-character-height-scaling, label-placement-window, label-durability-cycles, duplicate-harness-label-token, both-end-wire-marking."4license: Apache-2.05compliance: STANDARDS-REF6standards:7 - id: ecss8 reference-only: true9gated: false10domain: space-systems11pack: space-systems12compatibility: "agentskills.io SKILL.md; any SKILL.md host (Claude Code, Hermes, OpenClaw)"13metadata:14 source: https://skillmd.com/skills/ashfordeou/q2030-comp-labelling15 domain: space-systems16 subdomain: ecss17 tags: [ecss, q-st-20-electrical-harness-scope, q-st-20-30c, q2030-comp-labelling, complementary-harness-labelling, harness-label-token-format, marking-character-height-scaling, label-placement-window, duplicate-harness-label-token]18 version: 0.1.019 author: Aero Agent Skills20---2122# ECSS Harness Manufacturing — Complementary Labelling and Marking (space-systems/ecss/q2030-comp-labelling)2324Use when the task is the complementary labelling layer of ECSS-Q-ST-20-30C25clause 7.6 -- the space-specific demands on what a harness marking says,26how it is formed and how long it stays readable, on top of the workmanship27marking criteria the standard delegates to its adopted chapter.2829## Domain quick reference3031- A harness label is a traceability token, not a description. Its value32 comes from decomposing cleanly into the fields the project format33 prescribes, so a label that cannot be parsed has already failed however34 sensible it reads to a person standing at the bench.35- Content is graded field by field, because each position carries its own36 character set and length. A project code in lower case, a harness number37 without its prefix letter or a serial carrying a letter all parse and38 all still fail, and naming which field failed is what makes the finding39 actionable.40- Character height is not an absolute. It scales with the distance the41 marking has to be read from, so the same marking that is comfortable on42 a bench harness is illegible on an installed one behind a bracket, and43 the read distance belongs in the requirement rather than in the44 inspector's judgement.45- Placement is a window, not a preference. Too close to the termination46 and the marking is under the backshell or the strain relief; too far47 down the wire and it is lost inside the bundle at the first breakout.48- A wire that disappears into a bundle owes marking at both ends. One end49 identifies the wire at the connector it is being worked on; the far end50 is what makes a fault traceable without unlacing the harness.51- Durability is evidence, not a material claim. The marking survived a52 recorded number of abrasion and solvent exposures while staying legible,53 and those counts are what is graded against the project requirement.54- Uniqueness is the whole point. Two wires carrying the same token are55 worse than two unmarked wires, because the marking now actively56 misdirects the next person to trace them.5758## Workflow59601. Validate each label record: an identifier, the token, the character61 height, the read distance and the distance from its termination.622. Decompose the token against the prescribed field order and refuse a63 token whose field count does not match; an empty field is a refusal64 too, not an optional position.653. Grade every parsed field against the character set and length its66 position carries, naming the field in the finding.674. Compute the minimum character height from the read distance and grade68 the actual height against it, absorbing an exact equality with a69 tolerance instead of rounding the height up.705. Grade the distance from the termination against the placement window71 at both edges.726. Where the wire is routed through a bundle, confirm both ends carry the73 marking.747. Grade the abrasion and solvent legibility counts against the project75 requirement and report each shortfall with its margin.768. Collect every token in the harness, report any carried more than once,77 and roll all findings into one verdict.7879## Pitfalls8081- Accepting a label because a person could read it. The format is what82 makes the token machine-traceable through the as-built record; human83 legibility is necessary and nowhere near sufficient.84- Grading the token as one string. A single regular expression over the85 whole label tells the operator it is wrong without telling them which86 field to fix, and the rework comes back wrong twice.87- Using one minimum character height everywhere. The requirement follows88 the read distance, so an installed harness read from half a metre away89 needs taller marking than the same wire on the bench.90- Treating placement as tidiness. A label under the backshell cannot be91 read without demating, and a label past the first breakout is on the92 wrong side of it.93- Marking only the connector end of a bundled wire. The far end is the94 one a fault investigation needs, and adding it later means unlacing.95- Claiming durability from the marker material's datasheet. The96 requirement is graded on the legibility that survived recorded97 exposure, and a substitution of marker or solvent invalidates it.98- Letting a duplicate token through because both wires are in different99 connectors. The token is unique across the harness; scoping uniqueness100 to a connector is how two wires end up indistinguishable in the as-built101 record.102103## Behavior contract (gate 3)104105The token decomposition, the per-field content grading, the read-distance106character-height scaling, the placement-window grading, the both-end107marking rule, the abrasion and solvent durability verdict, the duplicate108token search and the whole-set roll-up are exercised by the gate 3109contract test: scripts/test_q2030_comp_labelling.py against110scripts/q2030_comp_labelling_logic.py (stdlib unittest, offline). Run:111python3 scripts/test_q2030_comp_labelling.py112113## Compliance114115- ECSS standards are freely downloadable (ESA); cite the source and116 paraphrase per standards-map.yaml.117- compliance: STANDARDS-REF, gated: false.