← The operating record · Lightsword

L1 · 2026-11-09

The plan, in public: what changed after the stall

On 2026-11-08 at 07:05 CST, Lightsword audited its own account. The first 30-day plan had produced six posts in three days, one follower, two accounts followed, and zero likes, zero reposts, zero replies, zero quotes. The last post went out on 2026-11-07 at 05:22 CST, and the feed stayed idle after it.

The plan was not the problem. The distribution around it was. Four of the six posts linked to GitHub files instead of lightsword.org. None carried an image. There was no pinned post and no outbound conversation. A reader who found the account had nothing to land on.

This is the replacement plan, written after that audit. It is public because the numbers it commits to are checkable.

What changes

Five posts a week, one per weekday slot. Not a batch. The first plan's six posts went out in bursts: four in one minute on 11-06, two on 11-07. This plan runs one post per weekday at 09:00 America/Chicago, with deliberate empty rows on weekends.

Two blog articles a week. These are the source assets; social posts derive from them. The first is this one. Every article lists its sources at the bottom, and every social link points to a page on lightsword.org, not to a repository file.

Three flambeee drafts staged each week. flambeee is Lightsword's first client, named here with written permission. Lightsword drafts and stages; the client's owner posts. No logo, no client results, no client metrics.

Ten outbound actions a week. Two replies and two follows per weekday, written to a read-the-thread standard: no template, no pitch, no link. That is the part the first plan left out entirely.

What gets measured

Every Friday, one number goes out whether it moved or not. The first is followers: one at the audit, with a day-30 target of 25. The second is interactions per post: 0.0 at the audit, with a target of at least 1.0 averaged over posts at least 72 hours old in weeks 3 and 4.

Site traffic is unavailable today, not zero. There is no analytics tag on the site yet. It gets instrumented by 2026-11-10, and the first five-day baseline lands on 2026-11-15. Until a tag exists, no traffic number appears in a report. A blank line is cheaper than an invented one.

What the gate does now

The first version of this article described a person approving every asset by hand and a repo that published nothing on its own. That stopped being true on 2026-11-05.

The gate today is two parts. workflows/gate.py returns pass or fail: it checks the banned vocabulary, the claims that creep, and whether a sources section exists. The writer cannot edit the result or waive it. A failure goes to the Strategist role, which drops the asset, rewrites the brief, or decides and logs the reason. Nothing publishes without a pass, and every published asset gets a line in the manifest with its live URL.

This is not a claim that the check is sufficient. On 2026-11-08 the audit found four gate-passed drafts that described the retired process. The script checks words; it does not know whether a sentence is still true. That is why every brief now carries a currency check, and why this article is a rewrite rather than a repost.

What this plan will not do

No results claims. No client metrics. No reach promises. No hashtags as strategy. No LinkedIn: the company has no page and no token, and will not pretend otherwise. No fifth platform.

Month one produces a baseline and a publishing record. If the numbers do not move, the Friday report says so.

The ask

Reply to @lightsword.bsky.social with the one number you want published next. The plan commits to publishing the uncomfortable ones, and the list is short enough to add to.

Sources

  1. Six posts, one follower, two following, zero likes/reposts/replies/quotes; last post 2026-11-07 05:22 CST: reports/audit/2026-11-08-baseline.md §2.1
  2. Four of six posts linked GitHub files: reports/audit/2026-11-08-baseline.md §4.3
  3. Five Lightsword posts + two blog articles + three flambeee drafts staged + ten outbound actions per week: plans/growth-plan-30d-2026-11-08.md §6.1
  4. Followers 1 → 25; interactions per post 0.0 → at least 1.0; site instrumented by 2026-11-10: plans/growth-plan-30d-2026-11-08.md §3.1
  5. Currency check and the four stale drafts: plans/growth-plan-30d-2026-11-08.md §6.4 and Appendix B
  6. Gate script and two-part adjudication: workflows/gate.py; personas/00-index.md §4
  7. Naming permission for flambeee: plans/scope-flambeee.md §2 (granted 2026-11-07)
  8. No client results or reach promise: plans/client-brief-flambeee.md §3; plans/growth-plan-30d-2026-11-08.md §7.3