Launch Tier Classification

Every release gets classified before planning begins. Source: Lauchengco (Loved).

Preflight: Read target canvas file(s) before any Write/Edit

Hard rule. Before issuing Write or Edit against any .claude/canvas/*.yml, use the Read tool on that file in this session. Claude Code's Read-before-Write check requires the Read tool specifically — cat/head/grep via Bash do NOT satisfy it.

Edit vs Write — different cost profiles (verified 2026-05-14):

• Edit (exact-string replacement): Read with limit: 1 satisfies the check at ~50 tokens. State-tracking is per-file, not per-byte — subsequent Edit calls work anywhere in the file. Use this for partial updates against large canvas files (e.g., purpose.yml at 800+ lines).
• Write (full replacement): do a full Read first. Write obliterates the file; you should see what you're about to replace. The limit:1 shortcut is not appropriate here.

ID-bearing entries — scan the ID space before assigning (added 2026-05-15, v0.23.19): When adding a new component, opportunity, solution, or any other ID-bearing entry to a canvas file, run a Bash grep first to confirm the next ID in your prefix sequence is actually free:

grep -o "<prefix>-[0-9][0-9]*" .claude/canvas/<file>.yml | sort -u -t- -k2 -n | tail -3

Replace <prefix> with the canvas's ID prefix (comp for landscape, opp for opportunities, sol for solutions, ht for human-tasks, etc.). Then pick the next free integer, matching the zero-padding already used in that file. The sort is NUMERIC (-t- -k2 -n) rather than lexical, and that is not pedantry: a plain sort -u orders ht-1 after ht-080, so on a canvas with inconsistent padding it reports the wrong maximum and the next ID collides. Verified on the dogfood repo 2026-08-13, where lexical sort returned ht-1 as the highest human-task ID against an actual ht-080. grep -o is also deliberate: it matches IDs wherever they appear, including cross-references and prose, so an ID that was promised somewhere but not yet defined is not handed out twice. validate_canvas.py has a duplicate-ID check (lines 230-239) that catches the failure on CI, but a duplicate can persist in the working tree for days if CI isn't run between edit and discovery — see roadmap-repo corrections.md 2026-05-15 "Duplicate canvas ID created in landscape.yml" for the worked example.

Original failure mode: anti-pattern #7 instance #5, 2026-05-09 — agent conflated Bash head with the Read tool, lost ~14k tokens to a Write-fail → remedial-full-Read → re-Write loop. The limit:1 discipline (graduated 2026-05-14, v0.23.18) prevents the second-order cost where the agent correctly follows the rule but full-Reads every time. The ID-scan discipline (graduated 2026-05-15, v0.23.19) prevents the related class where the agent reads enough of the file to satisfy the Edit check but not enough to see existing ID assignments — kin to anti-pattern #8 (Stale State Read).

If this skill writes to multiple canvas files, register each one first (limit:1 for Edit-only paths; full Read for Write paths) AND ID-scan any prefix you intend to assign.

See CLAUDE.md Canvas writes — Read before Write for the canonical rule.

Tier Definitions

Tier	Type	Effort	Examples
1	Major	Full cross-functional	New product, major pivot, category-defining
2	Significant	Targeted campaigns	Feature launch, positioning reinforcement
3	Incremental	Lightweight	Bug fixes, minor improvements, release notes

Classification Criteria

• Does this change our positioning? -> Tier 1
• Does this strengthen existing positioning? -> Tier 2
• Is this an incremental improvement? -> Tier 3

Per-Tier Activities

Software (default)

Tier 1: Press, events, campaigns, sales enablement, analyst briefings, customer advisory Tier 2: Blog post, targeted campaigns, sales enablement update, in-product announcement Tier 3: Release notes, changelog, in-product notification, knowledge base update

Content Products (courses, publications, media) (v0.11.0)

Tier 1: Platform launch (new course on marketplace), PR/media coverage, launch webinar, guest appearances Tier 2: New module/section, cross-promotion, community announcement, guest post Tier 3: Content update, errata fix, supplementary material, minor revision

AI Tools (v0.11.0)

Tier 1: Public launch, ProductHunt/HackerNews, documentation site, demo video Tier 2: New capability/model, integration partnership, case study Tier 3: Prompt improvement, model update, bug fix, eval result improvement

Service Offerings (v0.11.0)

Tier 1: New service line launch, case study PR, conference talk, partnership announcement Tier 2: New package/tier, testimonial campaign, process improvement announcement Tier 3: Pricing update, workflow refinement, expanded availability

Behavioral Science in Positioning (Shotton)

Use biases ETHICALLY to help users understand value:

• Social proof: Reference customers, usage numbers (real, not inflated)
• Anchoring: Frame value relative to alternatives
• Framing: Position the benefit, not just the feature
• Never: Confirmshaming, hidden costs, forced continuity, misdirection

Canvas Output

Update .claude/canvas/go-to-market.yml with tier classification and launch plan.

Ethical Engagement Design (Eyal -- Hook Model + Indistractable)

Eyal's work spans two complementary books: Hooked (2014) provides the Hook Model for building habit-forming products; Indistractable (2019) provides the user-side framework for managing attention. The Manipulation Matrix below bridges both — ethical engagement design means building hooks that users would choose even with full information.

The Hook Model is most relevant at L3 (Solution design) for engagement architecture, not just L5 (Market). Apply during solution design when the product requires recurring usage.

For products that need user retention, design engagement ethically using the Hook Canvas:

Hook Canvas

Map the four components of habit formation:

• Trigger: What prompts the user to engage? (External: notification, email. Internal: emotion, routine.)
• Action: What is the simplest behavior in anticipation of reward? (Must be easier than thinking.)
• Variable Reward: What reward satisfies the user's need while leaving them wanting more? (Tribe: social, Hunt: resources, Self: mastery.)
• Investment: What bit of work does the user put in that improves the next cycle? (Data, content, reputation, skill.)

Manipulation Matrix (Ethical Gate — NUDGE)

Before implementing engagement design, answer honestly:

1. Does it materially improve the user's life? (Not just "engagement" — actual value.)
2. Would you use it yourself? (The maker's test.)

	User Benefits	User Doesn't Benefit
Maker Uses It	Facilitator (ethical)	Entertainer (proceed with caution)
Maker Doesn't Use It	Peddler (risky)	Dealer (unethical — do not build)

Only Facilitator products should be built without reservation. Entertainers need honest self-assessment. Peddlers and Dealers trigger anti-pattern #10 (Dark Pattern Marketing).

Update .claude/canvas/go-to-market.yml engagement_design section with Hook Canvas results.

Source: Eyal (Hooked), with ethical framework from the Manipulation Matrix

Pre-Launch Bias Check

Before classifying a launch tier, run /mycelium:bias-check for L5-specific biases:

• Optimism bias: Are we overweighting positive signals and ignoring negative ones?
• Confirmation bias: Are we seeking validation that "it's ready to ship" rather than honestly assessing market readiness?
• Anchoring: Are we fixated on the initial positioning without considering what evidence now suggests?
• Sunk cost fallacy: Are we launching because we've invested too much to stop, not because the market signals are positive?

If /mycelium:bias-check reveals significant biases, address them before finalizing the launch tier.

After Launch: The L5 -> L2 Feedback Loop

This is critical. After launch, market feedback must flow back into discovery:

1. Capture market signals (within 2-4 weeks post-launch). Check the product-type-appropriate metrics canvas via /mycelium:dora-check:

   Software: feature usage, retention, conversion, support tickets, NPS/CSAT Content: refund rate, completion rate, drop-off points, return rate, reviews, NPS AI tool: task success rate, retention, DAU, refund rate, user feedback Service: client satisfaction (NPS/CSAT), referral rate, retention, delivery lead time feedback

2. Validate scenarios against reality (Hoskins):

   • For each scenario in .claude/canvas/scenarios.yml linked to this launch: did the persona's story play out?
   • Update lifecycle.validated_in_market: confirmed, partial, or invalidated
   • Invalidated scenarios are the most valuable learning — they reveal where the user model was wrong

3. Evaluate against L2 assumptions:

   • Do the signals confirm the L2 opportunity we solved for?
   • Are there NEW needs we didn't anticipate?
   • Did users "hire" the product for a different job than expected? (JTBD)
   • Do real user stories suggest NEW scenarios not in .claude/canvas/scenarios.yml?

4. Feed back into discovery:

   • If signals confirm: update confidence scores, mark scenarios as validated, celebrate validated learning
   • If signals reveal NEW opportunities: spawn a new L2 Opportunity diamond with market evidence as the starting data. Create new scenarios from real user stories.
   • If signals contradict: flag for diamond regression, mark scenarios as invalidated, update corrections.md

This closes the full Mycelium loop: Purpose -> Strategy -> Discovery -> Solution -> Delivery -> Market -> Discovery.

Cycle History Recording

After launch feedback is captured (L5 → L2 loop), update the cycle record in .claude/canvas/cycle-history.yml:

1. Find the cycle record for this leaf (created by /mycelium:retrospective at delivery completion)
2. Add actual market outcomes: user metrics, adoption data, NPS/CSAT, revenue impact
   • Source this data from /mycelium:metrics-pull where possible (v0.14): 24-48h after launch to capture the bump, then weekly for the first month. Snapshots live at .claude/evals/metrics/<source>/*.json. This replaces manual "I checked the dashboard" reports with timestamped evidence.
   • If .claude/jit-tooling/active-metrics.yml has no configured source for the relevant channel, run /mycelium:metrics-detect first.
3. Update the calibration section: compare predicted value/usability risk against actual market reception
4. If market signals contradict the original L2 opportunity assumptions, note this as calibration data

If no cycle record exists yet (leaf went directly to market without retrospective), create one now.

This closes the data loop: predicted ICE → actual delivery metrics → actual market outcomes → calibration for future scoring.

Decision Log (MANDATORY per G-P4)

APPEND a ### Launch Tier Classification entry to .claude/harness/decision-log.md with: tier assigned, positioning rationale, key risks, go-to-market approach.

Theory Citations

• Lauchengco: Loved (launch tier classification, positioning)
• Shotton: Choice Factory (ethical behavioral science in positioning)
• Kim: Three Ways (Second Way -- amplify feedback loops right-to-left)
• Torres: Continuous Discovery (market signals feed back into OST)

Counter-Argument Check (Bias Mitigation)

Before finalizing the launch tier classification AND before drafting any positioning copy, draft a one-line counter-argument: "What's the strongest case that this is a smaller tier than I'm claiming? That this positioning is overstating impact? That market readiness is weaker than the evidence suggests?" If you can't articulate one, run /mycelium:devils-advocate before proceeding.

This addresses the bias cluster documented in corrections.md (L5 sycophancy 2026-04-20 — promotional language in decision logs at L5 — explicitly named this skill's domain; eval overfitting 2026-04-30; sharper-framing-isn't-righter 2026-05-03). L5 work is the highest-risk context for bias because the framing pressure is explicit ("we're going to market"). G-M1 catches the WORST language, but bias appears in subtler tier-classification and positioning choices that G-M1 doesn't see. The counter-argument is the upstream check.

Postflight: Verify-After-Write (claim matches state)

Hard rule (per CLAUDE.md Communication Rules, anti-pattern #7 write-narration-verification — mechanism Check 42, graduated v0.39.18; enforced surface expanded to this skill v0.44.0). This skill mandates multi-field canvas updates. Before narrating "updated / wrote / refreshed [canvas]" in any user-facing summary, RE-READ the value fields this skill's MANDATORY says to update and confirm they actually changed — not just _meta.last_validated or a freshness stamp. Each field you claim to have updated must reflect its new value. The symmetric half of the Read-before-Write Preflight: that one protects what gets read before a write; this one protects that the write matches the claim. Worked failures: 2026-06-05 #18 (/dora-check narrated "updated" with value fields unchanged) + #19 (/retrospective left a cycle-history aggregate un-propagated).