Repository navigation
Conversation
Synter could reach Prospector's tools but had no playbook for them, so cold
email was the one motion the plugin could act on without guidance. These three
skills close that, and carry the corrections found auditing the live account on
2026-08-28.
**outbound** — the workflow skill. Pipeline, campaigns, inbox health, pause and
resume. Two things it insists on:
Roughly 90% of raw "replies" are out-of-office auto-responders. Measured
across the live account:
campaign | inbox replies | automated | human | interested
216 | 87 | 80 | 7 | 2
235 | 23 | 20 | 3 | 0
195 | 28 | 25 | 3 | 2
Campaign 216 read as "85 replies, 3.0%" while producing two interested
humans from 750 leads. The skill says to quote human replies and interested,
never the raw counter, and to say so when the honest figure has not synced
rather than substituting the inflated one.
And: never top up a campaign to fix a low reply rate. A campaign at 0.3% has
an offer problem; adding leads spends shared mailbox reputation to buy
nothing. That was very nearly done to campaign 216.
**outbound-research** — qualification. Disqualifiers (CRM, already met,
current customer, prior contact, suppression) veto rather than score, with a
lapsed customer as the deliberate win-back exception. Then a ranked signal
hierarchy with decay windows, where role-plus-industry is explicitly not a
signal. Then how to state the fact without naming the surveillance, the banned
openers that read as generated, day-of-week anchors (including why "Happy
Monday!" signals bulk mail), the fiscal calendar read from the prospect's year
rather than ours, and life-event hooks with their windows.
Its test is the operator one: for a real lead, if they asked "where did you
get my information?", what would you say? That is a verbatim reply from a
campaign that mailed 750 people an identical email.
**outbound-replies** — triage. Strips auto-responders, our own Sent copies and
bounces before counting, classifies what remains, routes opt-outs straight to
suppression, and holds every draft for explicit approval. Approval for one
reply never carries to the next message in the thread.
Skills auto-discover, but the README enumerates them, so both listings are
updated.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01D6Vx5DgLaKF3E9gw9xUn3q
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Synter can already reach Prospector's tools (
prospector_get_campaigns,prospector_create_campaign,prospector_create_reply, …) but had no playbook for them — cold email was the one motion the plugin could act on without guidance. These three skills close that, and carry the corrections found auditing the live account on 2026-08-28.outbound— the workflow skillPipeline, campaigns, inbox health, pause and resume. Two things it insists on.
Auto-responders are not replies. Roughly 90% of the raw count is out-of-office:
Campaign 216 read as "85 replies, 3.0%" while producing two interested humans from 750 leads. The skill says to quote human replies and interested, never the raw counter, and to say so when the honest figure hasn't synced rather than substituting the inflated one.
Never top up a campaign to fix a low reply rate. A campaign at 0.3% has an offer problem; adding leads spends shared mailbox reputation to buy nothing. That was very nearly done to 216.
outbound-research— qualificationDisqualifiers (CRM, already met, current customer, prior contact, suppression) veto rather than score — a lapsed customer being the deliberate win-back exception. Then a ranked signal hierarchy with decay windows, where role-plus-industry is explicitly not a signal.
Then the craft: how to state a fact without naming the surveillance ("I noticed you visited our pricing page" fails the conference test), the banned openers that read as generated, day-of-week anchors including why "Happy Monday!" signals bulk mail, the fiscal calendar read from the prospect's year rather than ours, and life-event hooks with their windows.
Its test is the operator one: for a real lead, if they asked "where did you get my information?" — what would you say? That is a verbatim reply from a campaign that mailed 750 people an identical email.
outbound-replies— triageStrips auto-responders, our own Sent copies (outbound returns through the same feed) and bounces before counting. Classifies what remains, routes opt-outs straight to suppression, and holds every draft for explicit approval — approval for one reply never carries to the next message in the thread.
Notes
Skills auto-discover, but the README enumerates them, so the workflow table and craft-skill listing are both updated. Docs only — no code, no behavior change.
Companion PR in
synter-prospector-aiadds the deeper repo-side version for engineers working in that codebase: Synter-Media-AI/synter-prospector-ai#262🤖 Generated with Claude Code
https://claude.ai/code/session_01D6Vx5DgLaKF3E9gw9xUn3q