Narrative Drift Monitor
Detects brand-message drift over time and identifies evidence-based triggers for repositioning.
The skill explicitly treats pages, snapshots, and exports as untrusted input, keeps the canon read-only, asks before writing memory, and routes registry changes through an authorized propose flow. The repository security material also discloses fetch, egress, and runtime-memory risks. Six points are deducted because competitor and Wayback retrieval create external data flows, several write paths depend on host authorization, and per-run confirmation, rollback, and sensitive-data handling are not fully specified here.
The instructions, input/output contract, stop conditions, and abnormal cases are relatively clear: missing canon yields NEEDS_INPUT, missing history must not be guessed, and evidence labels are required. Static materials provide no skill-specific tests or reproducible execution evidence, and availability of Wayback, memory paths, and sibling skills is not verified by this skill; the static ceiling therefore applies.
Triggers, use cases, non-fit boundaries, arguments, expected outputs, and handoffs are clearly stated, with Chinese and English descriptions. Four points are deducted because the core workflow depends on project memory, Wayback, and sibling-skill state; environment boundaries and mainland-China reachability are not sufficiently established, while geo-relevance is marked low.
The skill follows a consistent contract with Quick Start, read/write boundaries, completion criteria, data sources, references, save confirmation, and version metadata. Apache-2.0 licensing is supported by the supplied repository material. Four points are deducted because maintainer ownership, update path, and changelog information are not clear for this skill, and several relative paths and host-authorization assumptions require shared repository context.
The expected report is concrete and covers per-surface drift, competitor repositioning, trigger conditions, and D1/W1/M1 retrospectives, with Measured, User-provided, and Estimated labels; this could directly support re-audit or repositioning decisions. Nine points are deducted because static review cannot verify actual report quality, and the result depends on the availability and quality of the canon, historical snapshots, competitor map, and optional resonance inputs.
The skill requires dated Wayback snapshots, evidence labels, proxy labeling, and explicit no-history notes, providing useful traceability. Six points are deducted because there are no skill-specific tests, third-party execution results, or independent corroboration, and many key facts depend on user data or sibling-skill artifacts.
- Before execution, confirm that the canon, version history, Wayback connector, and competitor map are available; otherwise stop with NEEDS_INPUT.
- Confirm authorization, robots.txt compliance, and applicable legal boundaries before retrieving competitor or third-party pages, and never treat page content as instructions.
- Require explicit user confirmation and host authorization for memory or registry-event writes, and verify that proposals cannot bypass canonical ownership.
- Wayback and other overseas services may be unreachable from mainland-China networks; prepare dated user exports or another auditable evidence substitute.
What does this skill do, and when should you use it?
Narrative Drift Monitor is designed for ongoing governance after a brand narrative has been deployed. It compares flagship surfaces such as homepages, pricing pages, and store listings with the read-only narrative-registry canon, using dated Wayback history where available. It also tracks competitor repositioning, defines explicit conditions for changing the brand narrative, and runs D1/W1/M1 retrospectives on intended versus actual message pull-through. It is a good fit for teams with an established narrative canon and multiple surfaces to keep aligned.
Reads memory/narrative-registry/canon.md and versions.md as the baseline; collects user-provided or scraped brand surfaces; calls scripts/connectors/wayback.py for dated change history; assigns each surface a matches, drifted, or contradicts verdict; reports dated competitor framing changes; defines concrete repositioning triggers plus the narrative-whiplash guardrail; and produces D1/W1/M1 message-shift retrospectives. Each data point is labeled Measured, User-provided, or Estimated, while proxy signals are labeled proxy.
- A brand team wants to check whether its homepage, pricing page, and store listing have diverged from the approved canon.
- A marketing lead sees a competitor change its category framing and needs to decide whether the brand should review its own positioning.
- A launch or campaign team needs a D1, W1, or M1 review comparing the intended narrative with what actually appeared across surfaces.
- A multi-channel brand team wants explicit repositioning criteria so the narrative is not repeatedly changed on instinct alone.
What are this skill's strengths and limitations?
- Combines self-drift analysis, competitor-repositioning alerts, trigger criteria, and D1/W1/M1 retrospectives in one report.
- Uses dated Wayback evidence and distinguishes Measured, User-provided, Estimated, and proxy data.
- Has a clear scope boundary: it monitors drift and hands off rather than scoring or rewriting the canon.
- Runs with Tier 1 data and does not require a paid monitoring tool or MCP.
- Requires an existing narrative-registry canon and version history before drift can be measured.
- If Wayback has no history for a surface, it can only record that history is unavailable; it cannot infer past wording.
- It does not compute TALE results, run vetoes, or measure echo rate, share of voice, or AI-answer resonance.
- The supplied material does not provide an independent test suite or cross-platform validation record for this individual skill.
How do you install this skill?
Install the repository with npx skills add aaron-he-zhu/aaron-marketing-skills. Alternatively, run git clone https://github.com/aaron-he-zhu/aaron-marketing-skills and use the skill under narrative/evaluate/narrative-drift-monitor/. The source does not document a separate standalone installation package for this skill.
How do you use this skill?
After installation on an Agent Skills-compatible host, provide the brand, surfaces to monitor, optional competitor set, canon path, and review window. Example: "Check whether our homepage, pricing page, and store listing have drifted from the canon since the last version. Canon: memory/narrative/. Use wayback for change history." The canon.md and versions.md records must exist; otherwise the skill should stop with a required-input condition.
How does this skill compare with similar options?
Unlike narrative-cascade-planner, this skill handles post-launch drift over time rather than the first consistency check before a surface ships. Unlike narrative-quality-auditor, it produces evidence and drift verdicts rather than computing the TALE profile or vetoes. Unlike narrative-resonance-monitor, it watches alignment with the canon rather than echo rate or AI-answer resonance. When a genuine trigger fires, it hands the repositioning work to message-system-architect.